ARTIFICIAL

INTELLIGANCE

Skill Enhance Course

(As per National Education Policy 2020)

1
Bangalore University

ABOUT AUTHORS:

Prof. RESHMA B
B.E, MTch.
Assistant Professor and Head,
BCA Department
Seshadripuram Academy of Business studies
Kengeri, Bangalore - 560060

Prof. AKSHATHA M.R

B.E, MTch.
Assistant Professor, BCA Department
Seshadripuram Academy of Business studies
Kengeri, Bangalore - 560060

Prof. AMABRISH R
M.Com, MBA, Dip. IFRS, UGC-NET
Principal, Dharmasagara First Grade College,
Dommasandra, Anekal Taluk,
Bangalore - 262125.

Prof. DADAVALI S.P

MCA, Mphil,
Assistant Professor,
Government First Grade College,
Kengeri, Bangalore - 560060

2
 PREFACE

At least since the first century BCE, humans have been intrigued by the
possibility of creating machines that mimic the human brain. In modern times, the
term artificial intelligence was coined in 1955 by John McCarthy. In 1956,
McCarthy and others organized a conference titled the “Dartmouth Summer
Research Project on Artificial Intelligence.” This beginning led to the creation of
machine learning, deep learning, predictive analytics, and now to prescriptive
analytics. It also gave rise to a whole new field of study, data science.

Today, the amount of data that is generated, by both humans and machines,
far outpaces humans’ ability to absorb, interpret, and make complex decisions
based on that data. Artificial intelligence forms the basis for all computer learning
and is the future of all complex decision making. As an example, most humans can
figure out how to not lose at tic-tac-toe (noughts and crosses), even though there
are 255,168 unique moves, of which 46,080 end in a draw. Far fewer folks would
be considered grand champions of checkers, with more than 500 x 1018, or 500
quintillion, different potential moves. Computers are extremely efficient at
calculating these combinations and permutations to arrive at the best decision. AI
(and its logical evolution of machine learning) and deep learning are the
foundational

Finally we profusely thank Branch Managers of Kalyani Publishers Mr.

Mohammed Sameer for coming forward to publish our work. We will accept the
comments and suggestion from our readers to improve the contents.

Prof. RESHMA B
Prof. AKSHATHA M.R
Prof. AMABRISH R
Prof. DADAVALI S.P

3
 ACKNOWLEDGEMENT

We are thankful to a large number of intellectual individuals who have

encouraged us in authoring this book. We would like to express our deep sense of
gratitude to authors and scholars in the field of management who have shaped us
in understanding through their rich and valuable contributions and also to our
students fraternity for providing us the stimulus needed for shaping this book in a
useful manner.

In particular we wish to thank profusely Dr. Ramachandra Gowda Vice

Chancellor Rani Chennamma University, Dr.R.Chenraj Jain, Chairman Jain Group of
Institutions, Dr. N. Sundararajan, Vice Chancellor, Jain University, Prof. V.H.
Rajshekhar Principal GFGC Ramanagara Dr.B.T.Venkatesh, Principal, Jain
University, Dr. B.Anuradha Principal, Al-Ameen Institute of Management Studies
Bangalore; Mr.Ravikumar S of Eshcol Cybervilla, Kengeri, Bangalore for their
invaluable encouragement, Guidance and Contributions in shaping this book.

Finally we express our deep sense of gratitude of our family members for
their whole hearted support in authoring this book in a useful manner.

Prof. RESHMA B
Prof. AKSHATHA M.R
Prof. AMABRISH R
Prof. DADAVALI S.P

4
 SYLLABUS

Course Content (Artificial Intelligence)

Details of topic Duration
Course – 1 - AI-900 pathway consists of 5 courses and 2 05 hours
Azure AI reading material:
Fundamentals i. Introduction to AI on Azure
(AI-900) ii. Use visual tools to create machine learning
models with
Azure Machine Learning
iii. Explore computer vision in Microsoft
Azure
iv. Explore natural language processing
v. Explore conversational AI
vi. Tune Model Hyperparameters - Azure
Machine Learning
(Reading)
vii. Neural Network Regression: Module
Reference - Azure
Machine Learning (Reading
Practical 1. Prepare the data 13 hours
2. Model the data
3. Visualize the data
4. Analyse the data
5. Deploy and maintain deliverables
Course – 2 - DA-100 pathway consists of 5 courses and 2 08 hours
Data Analyst reading material:
Associate 1. Get started with Microsoft data
(DA-100) analytics
2. Prepare data for analysis
3. Model data in Power BI
4. Visualize data in Power BI
5. Data analysis in Power BI
6. Manage workspaces and datasets in
Power BI
7. Key Influencers Visualizations Tutorial
- Power BI
8. Smart Narratives Tutorial - Power BI
Microsoft Docs
Practical 1. Describe Artificial Intelligence 13 hours
workloads and considerations

5
 2. Describe fundamental principles of
machine learning on Azure
3. Describe features of computer vision
workloads on Azure
4. Describe features of Natural Language
Processing (NLP) workloads on Azure
Course Outcomes (COs):

• Appraise the theory of Artificial intelligence and list the significance

of AI.
• Discuss the various components that are involved in solving an AI
problem.
• Illustrate the working of AI Algorithms in the given contrast.
• Analyze the various knowledge representation schemes, Reasoning
and Learning techniques of AI.
• Apply the AI concepts to build an expert system to solve the real-
world problems.

6
 COURSE – 1

AZURE AI FUNDAMENTALS (AI-900)

1.1 INTRODUCTION:

AI is touching us in all aspects of our daily lives, most of the times

unknowingly. Whenever we shop online, use our mobiles, drive to work
daily, check our mail box or exercise, AI is coming into play and helping
us, prodding us or controlling us. Since AI is already such an integral
part of our lives, it makes sense to get more knowledge of this emerging
technology. From chess-playing computers to self-driving vehicles,
Artificial Intelligence (AI) is progressing rapidly and touching every aspect
of our lives. In this module, you will learn how machines can be made to
learn from data and carry out human tasks. AI enables us to build
amazing software that can improve health care, enable people to
overcome physical disadvantages, empower smart infrastructure, create
incredible entertainment experiences, and even save the planet!

1.2 WHAT IS ARTIFICIAL INTELLIGENCE?

Artificial intelligence is a wide-ranging branch of computer science

concerned with building smart machines capable of performing tasks
that typically require human intelligence. Simply put, AI is the creation
of software that imitates human behaviours and capabilities.

7
1.2.1 ARTIFICIAL INTELLIGENCE DEFINITION: BASICS OF AI

AI is intelligence demonstrated by machines, as opposed to the natural

intelligence displayed by humans or animals.

Key workloads include:

• Machine learning - This is often the foundation for an AI system, and

is the way we "teach" a computer model to make prediction and draw
conclusions from data.
• Anomaly detection - The capability to automatically detect errors or
unusual activity in a system.
• Computer vision - The capability of software to interpret the world
visually through cameras, video, and images.
• Natural language processing - The capability for a computer to
interpret written or spoken language, and respond in kind.
• Knowledge mining - The capability to extract information from large
volumes of often unstructured data to create a searchable knowledge
store.

8
1.3 Use visual tools to create machine learning models with Azure
Machine Learning:

The phrase “Every model is wrong, but some are useful” is especially true
in Machine Learning. When developing machine learning models, you
should always understand where it works as expected and where it fails
miserably.

An important method of understanding a model is through model

visualization. Visualizing the model architecture is important for:

• Model explain ability

• Results interpretation
• Model debugging
• Model comparison

Once you get some decent understanding of one model, you are good,
right? No, it is wrong. !!!. Typically, you need to do some or a lot of
experimenting with model improvement ideas, and visualizing differences
between various ML experiments becomes crucial.

There are many methods that you can use to get that understanding:

• Look at evaluation metrics (also, you should know how to choose an

evaluation metric for your problem)
• Look at performance charts like ROC, Lift Curve, Confusion Matrix,
and others
• Look at learning curves to estimate overfitting
• Look at model predictions on best/worst cases
• Look how resource-intensive is model training and inference (they
translate to serious costs and will be crucial to the business side of
things)
• Use model interpretation tools and techniques to vet predictions

9
1.3.1 Tools for Machine Learning experiments visualization

1. Neptune: Neptune is a metadata store for MLOps, built for teams that
run a lot of experiments. It gives you a single place to log, store, display,
organize, compare, and query all your model-building metadata.
Neptune is used for:

• Experiment tracking: Log, display, organize, and compare ML

experiments in a single place.
• Model registry: Version, store, manage, and query trained models
and model building metadata.
• Monitoring ML runs live: Record and monitor model training,
evaluation, or production runs live.

How can Neptune help you visualize experiments and models?

• Log any metadata type, including parameters and metrics, but also rich
objects like images, video, audio, and interactive visualizations.
• Visualize logged metadata and analyze results in a preferred way – in
the runs table, as charts, in dashboards, or in a folder structure.
• Compare hyper parameters and metrics across many runs with an
intelligent compare table that highlights what was different.
• See how different parameters and con-figs affect the results, and debug
and optimize models.
• Automatically monitor hardware utilization (GPU, CPU, memory).
• Register models and metadata associated with them.

2. Weights & Biases: Weights & Biases is a machine learning platform for
developers to build better models faster. It lets you quickly track
experiments, version and iterate on datasets, evaluate model
performance, reproduce models, visualize results and spot regressions,
and share findings with colleagues.

How can Weights & Biases help you visualize experiments and models?

10
• Monitor training runs information like loss, accuracy (learning curves)
• View histograms of weights and biases (no pun intended), or gradients
• Log rich objects like, charts, video, audio or interactive charts during
training
• Use various comparison tools like tables showing auto-diffs, parallel
coordinates plot and others
• Interactive prediction bounding box visualization for object detection
models
• Interactive prediction masks visualization for semantic segmentation
models
• Visualize live metrics like GPU and CPU utilization
• Build dataset dependency graphs
• Visualize parameter importance

3. Comet: Comet is an ML platform that helps data scientists track,

compare, explain and optimize experiments and models across the
model’s entire lifecycle, i.e. from training to production. In terms of
experiment tracking, data scientists can register datasets, code
changes, experimentation history, and models. Comet is available for
teams, individuals, academics, organizations, and anyone who wants
to easily visualize experiments, facilitate work, and run experiments. It
can be used as a hosted platform or deployed on-premise.

Main advantages:

• Fully-customizable experiment table within the web-based user

interface;
• Extensive comparison features—code, hyperparameters, metrics,
predictions, dependencies, system metrics, and more
• Dedicated modules for vision, audio, text, and tabular data that
allow for easy identification of issues with the dataset.

11
4. Sacred + Omniboard: Sacred is open-source software that allows
machine learning researchers to configure, organize, log, and reproduce
experiments. Sacred does not come with its proper UI but there are a
few dashboarding tools that you can connect to it, such as Omniboard
(but you can also use others, such as Sacred board, or Neptune via
integration). Sacred doesn’t have the scalability of the previous tools
and has not been adapted to team collaboration (unless integrated with
another tool), however, it has great potential when it comes to individual
research.

Main advantages:

• Possibility to connect it to the preferred UI;

• Possibility to track any model training developed with any Python
library;
• Extensive experiment parameters customization options.

5. MLflow: MLflow is an open-source platform that helps manage the

whole machine learning lifecycle. This includes experimentation, but
also model storage, reproducibility, and deployment. Each of these four
elements is represented by one MLflow component: Tracking, Model
Registry, Projects, and Models. The MLflow Tracking component
consists of an API and UI that support logging various metadata
(including parameters, code versions, metrics, and output files) and
later visualizing the results.

Main advantages:

• Focus on the whole lifecycle of the machine learning process;

• Strong and big community of users that provide community support;
• Open interface that can be integrated with any ML library or
language.

12
1.4 Explore Computer vision

Computer vision is a field of artificial intelligence (AI) that enables

computers and systems to derive meaningful information from digital
images, videos and other visual inputs - and take actions or make
recommendations based on that information. If AI enables computers to
think, computer vision enables them to see, observe and understand.
Computer vision works much the same as human vision, except humans
have a head start. Human sight has the advantage of lifetimes of context
to train how to tell objects apart, how far away they are, whether they are
moving and whether there is something wrong in an image. Computer
vision trains machines to perform these functions, but it must do it in
much less time with cameras, data and algorithms rather than retinas,
optic nerves and a visual cortex. Because a system trained to inspect
products or watch a production asset can analyse thousands of products
or processes a minute, noticing imperceptible defects or issues, it can
quickly surpass human capabilities. Computer vision is used in
industries ranging from energy and utilities to manufacturing and
automotive – and the market is continuing to grow. It is expected to reach
USD 48.6 billion by 2022.

1.4.1 How does computer vision work?

Computer vision is an area of artificial intelligence (AI) in which software

systems are designed to perceive the world visually, though cameras,
images, and video. There are multiple specific types of computer vision
problem that AI engineers and data scientists can solve using a mix of
custom machine learning models and platform-as-a-service (PaaS)
solutions - including many cognitive services in Microsoft Azure.

Computer vision is one of the core areas of artificial intelligence (AI), and
focuses on creating solutions that enable AI applications to "see" the
world and make sense of it.

13
 Of course, computers don't have biological eyes that work the way ours
do, but they are capable of processing images; either from a live camera
feed or from digital photographs or videos. This ability to process images
is the key to creating software that can emulate human visual
perception.

Some potential uses for computer vision include:

• Content Organization: Identify people or objects in photos and

organize them based on that identification. Photo recognition
applications like this are commonly used in photo storage and social
media applications.
• Text Extraction: Analyze images and PDF documents that contain
text and extract the text into a structured format.
• Spatial Analysis: Identify people or objects, such as cars, in a space
and map their movement within that space.

To an AI application, an image is just an array of pixel values. These

numeric values can be used as features to train machine learning models that
make predictions about the image and its contents.

Training machine learning models from scratch can be very time intensive
and require a large amount of data. Microsoft's Computer Vision service gives

14
you access to pre-trained computer vision capabilities.

Computer Vision models and capabilities

Most computer vision solutions are based on machine learning models

that can be applied to visual input from cameras, videos, or images. The
following table describes common computer vision tasks.

▪ Image classification: Image classification involves training a machine

learning model to classify images based on their contents. For example, in
traffic monitoring solution you might use an image classification model to
classify images based on the type of vehicle they contain, such as taxis,
buses, cyclists, and so on.
▪ Object detection: Object detection machine learning models are trained
to classify individual objects within an image, and identify their location
with a bounding box. For example, a traffic monitoring solution might use
object detection to identify the location of different classes of vehicle.
▪ Semantic segmentation: Semantic segmentation is an advanced machine
learning technique in which individual pixels in the image are classified
according to the object to which they belong. For example, a traffic
monitoring solution might overlay traffic images with "mask" layers to
highlight different vehicles using specific colors.

15
▪ Image analysis: You can create solutions that combine machine learning
models with advanced image analysis techniques to extract information
from images, including "tags" that could help catalogue the image or even
descriptive captions that summarize the scene shown in the image.

▪ Face detection, analysis, and recognition: Face detection is a specialized

form of object detection that locates human faces in an image. This can be
combined with classification and facial geometry analysis techniques to
recognize individuals based on their facial features.
▪ Optical character recognition (OCR): Optical character recognition is a
technique used to detect and read text in images. You can use OCR to read

16
 text in photographs (for example, road signs or store fronts) or to extract
information from scanned documents such as letters, invoices, or forms.

1.4.2 Explore Natural Language Processing

Natural language processing supports applications that can see, hear,

speak with, and understand users. Using text analytics, translation,
and language understanding services, Microsoft Azure makes it easy to
build applications that support natural language.

Analyzing text is a process where you evaluate different aspects of a

document or phrase, in order to gain insights into the content of that
text. For the most part, humans can read some text and understand
the meaning behind it. Even without considering grammar rules for the
language the text is written in, specific insights can be identified in the
text.

As an example, you might read some text and identify some key phrases
that indicate the main talking points of the text. You might also
recognize names of people or well-known landmarks such as the Eiffel
Tower. Although difficult at times, you might also be able to get a sense
for how the person was feeling when they wrote the text, also commonly
known as sentiment.

Text Analytics Techniques

Text analytics is a process where an artificial intelligence (AI) algorithm,

running on a computer, evaluates these same attributes in text, to determine
specific insights. A person will typically rely on their own experiences and
knowledge to achieve the insights. A computer must be provided with similar
knowledge to be able to perform the task. There are some commonly used
techniques that can be used to build software to analyze text, including:

17
 • Statistical analysis of terms used in the text. For example, removing
common "stop words" (words like "the" or "a", which reveal little
semantic information about the text), and performing frequency
analysis of the remaining words (counting how often each word appears)
can provide clues about the main subject of the text.
• Extending frequency analysis to multi-term phrases, commonly known
as N-grams (a two-word phrase is a bi-gram, a three-word phrase is a
tri-gram, and so on).
• Applying stemming or lemmatization algorithms to normalize words
before counting them - for example, so that words like "power",
"powered", and "powerful" are interpreted as being the same word.
• Applying linguistic structure rules to analyze sentences - for example,
breaking down sentences into tree-like structures such as a noun
phrase, which itself contains nouns, verbs, adjectives, and so on.
• Encoding words or terms as numeric features that can be used to train
a machine learning model. For example, to classify a text document
based on the terms it contains. This technique is often used to perform
sentiment analysis, in which a document is classified as positive or
negative.
• Creating vectorized models that capture semantic relationships
between words by assigning them to locations in n-dimensional space.
This modeling technique might, for example, assign values to the words
"flower" and "plant" that locate them close to one another, while
"skateboard" might be given a value that positions it much further
away.

While these techniques can be used to great effect, programming them

can be complex. In Microsoft Azure, the Language cognitive service can help
simplify application development by using pre-trained models that can:

• Determine the language of a document or text (for example, French or

English).
• Perform sentiment analysis on text to determine a positive or negative
sentiment.

18
 • Extract key phrases from text that might indicate its main talking
points.
• Identify and categorize entities in the text. Entities can be people,
places, organizations, or even everyday items such as dates, times,
quantities, and so on.

1.4.3 Explore conversational AI

Conversational AI is an umbrella term used to describe various methods

of enabling computers to carry on a conversation with a human. This
technology ranges from fairly simple natural language processing (NLP)
to more sophisticated machine learning (ML) models that can interpret
a much wider range of inputs and carry on more complex conversations.

One of the most common applications of conversational AI is in chat

bots, which use NLP to interpret user inputs and carry on a
conversation. Other applications include virtual assistants, customer
service chatbots, and voice assistants.

Savvy consumers expect to communicate via mobile app, web,

interactive voice response (IVR), chat, or messaging channels. They look
for a consistent and enjoyable experience that’s fast, easy, and
personalized.

For businesses, the key to meeting and exceeding these expectations

across channels and at scale is intelligent automation. Conversational
artificial intelligence (AI) powers interactions that are near human,
improving CX, boosting satisfaction, driving loyalty, and increasing
customer lifetime value (LTV).

1.4.4.1 Components of Conversational AI

Conversational AI can be broken down into five core components. These

five core components work together to enable a computer to understand and
respond to human conversation:

1. Natural language processing: NLP is the ability of a computer to

19
 understand human language and respond in a way that is natural for
humans. This involves understanding the meaning of words and the
structure of sentences, as well as being able to handle idiomatic
expressions and slang.

NLP is made possible by machine learning, which is used to train

computers to understand language. NLP algorithms use large data sets to
learn how words are related to each other, and how they are used in
different contexts.

2. Machine learning: Machine learning is a field of artificial intelligence that

enables computers to learn from data without being explicitly
programmed. Machine learning algorithms can automatically improve
their performance as they are exposed to more data.

Machine learning is used to train computers to understand language, as

well as to recognize patterns in data. It is also used to create models of how
different things work, including the human brain.

3. Text analysis: Text analysis is the process of extracting information from

text data. This involves identifying the different parts of a sentence, such
as the subject, verb, and object. It also includes identifying the different
types of words in a sentence, such as nouns, verbs, and adjectives.

Text analysis is used to understand the meaning of a sentence, as well as

the relationships between different words. It is also used to identify the
topic of a text, as well as the sentiment (positive or negative) of the text.

4. Computer vision: Computer vision is the ability of a computer to interpret

and understand digital images. This involves identifying the different
objects in an image, as well as the location and orientation of those objects.
Computer vision is used to identify the contents of an image, as well as the
relationships between different objects in the image. It is also used to
interpret the emotions of people in photos, and to understand the context
of a photo.
5. Speech recognition: Speech recognition is the ability of a computer to

20
 understand human speech. This involves recognizing the different sounds
in a spoken sentence, as well as the grammar and syntax of the sentence.

Speech recognition is used to convert spoken words into text, and to

understand the meaning of the words. It is also used to interpret the
emotions of people speaking in a video, and to understand the context of
a conversation.

1.4.4.2 How Does Conversational AI Work?

Driven by underlying machine learning and deep neural networks (DNN),

a typical conversational AI flow includes:

❖ An interface that allows the user to input text into the system or
Automatic Speech Recognition (ASR), a user interface that converts
speech into text.
❖ Natural language processing (NLP) to extract the user's intent from the
text or audio input, and translate the text into structured data.
❖ Natural Language Understanding (NLU) to process the data based on
grammar, meaning, and context; to comprehend intent and entity; and
to act as a dialogue management unit for building appropriate
responses.
❖ An AI model that predicts the best response for the user based on the
user's intent and the AI model's training data. Natural Language
Generation (NLG) infers from the above processes, and forms an
appropriate response to interact with humans.

In many cases, the user interface, NLP, and AI model are all provided
by the same provider, often a conversational AI platform provider. However,
it's is also possible to use different providers for each of these components.

21
1.4.4.3 How to create Conversational AI?

There is no one-size-fits-all answer to this question, as the best way to

create conversational AI depends on the specific needs and use cases of your
organization. However, some tips on how to create conversational AI include:

1. Start by understanding your use cases and requirements: The first step
in creating conversational AI understands your organization’s specific
needs and use cases. What are you trying to achieve with your chatbot?
What type of conversations do you want it to be able to have? What data
do you need to collect and track? Defining these requirements will help
you determine the best approach to creating your chatbot.
2. Choose the right platform and toolkit: There are several different
platforms and toolkits that you can use to create conversational AI. Each
platform has its own strengths and weaknesses, so you need to choose the
platform that best suits your needs. Some popular platforms include
[24]7.ai Conversations, Microsoft Bot Framework, Amazon Lex, Google
Dialogflow, and IBM Watson
3. Build a prototype: Once you have defined your requirements and chosen
a platform, it’s time to start building your prototype. Building a prototype
will help you test your chatbot and iron out any kinks before deploying it
to your users.
4. Deploy and test your chatbot: Once your prototype is finished, it’s time

22
 to deploy and test your chatbot. Make sure to test it with a small group of
users first to get feedback and make any necessary adjustments.
5. Optimize and improve your chatbot.: The final step is to continually
optimize and improve your chatbot. You can do this by tweaking the
algorithms, adding new features, and collecting user feedback.

Difference between Traditional chatbots and AI – Powered Chatbots:

Traditional AI-powered chatbots

Chatbots
Low complexity Focused, transactional Complex, contextual

▪ Basic answer and ▪ Can manage complex ▪ Goes beyond

response dialogues conversations
machines ▪ Integrate with ▪ Contextually aware
▪ Allow for simple multiple and intelligent
integration legacy/back-end ▪ Can self-learn and
▪ Based on limited systems improve over time

23
 scope ▪ Based on large scope ▪ Can anticipate user
▪ Need explicit ▪ Sepecilize in needs
training for every completing tasks ▪ Require massive
scenario interacting with back-end effort
▪ Require low multiple systems
back-end effort ▪ Require high back-
end effort

Conversational AI Challenges

Conversational AI’s maturity has steadily increased over the past few
years to the point where it can now deliver excellent business value and
outcomes for companies. Nevertheless, challenges abound since this is also a
fast-evolving conversational commerce category where very few vendors are
constantly innovating and bringing new technologies to the market.
Challenges include:

• Developing natural language processing (NLP) capabilities that can

understand and interpret human interactions. This is a complex task
that requires significant effort and investment in research and
development.
• Understanding the context of a conversation in order to provide
accurate responses. This can be particularly challenging in
conversations that involve multiple people or multiple topics.
• The need for sophisticated design and development efforts to create a
customer experience that engages users and keeps them engaged in the
conversation.
• Deploying and integrating a Conversational AI solution into an existing
business or application can be a significant challenge. Proper planning
and execution are essential to ensure a successful deployment.
• Ensuring the security and privacy of data exchanged via multiple
conversational AI-powered channels—this applies to any CX-related
information exchange. Compliance with standards such as GDPR,
CCCP, and other country-specific regulations is also critical.

24
 • As conversational AI permeates global CX platforms, local language
support becomes a high priority. Leading brands operating worldwide
can’t rely on availability in just one language to meet local needs at
scale. Building a robust conversational AI platform to operate in
regional languages, dialects, slang, noisy environments, with crosstalk,
etc., is a huge challenge.
• Dialogue management and conversation design are non-trivial parts of
conversational AI. Annotating the intelligence gathered from real agent
conversations and building the right model-training data requires
ongoing human-in-the-loop expertise.
• Building a conversational AI-based application that takes into
consideration intent, entity extraction, sentiment analysis, and
empathy is challenging and very few vendors offer solutions with these
features.
• Explainable AI—not all conversational AI platforms use this data
science tool, which eliminates algorithmic black boxes and helps
answer the “why” within the model’s functionality. Explainable AI also
improves trust in the platform’s ability to produce accurate, fair, and
transparent results.
• Keeping automated conversations relevant can also be a real challenge,
with customer needs and preferences changing faster than ever before.
As a result, you may need people with coding skills, multiple-persona
models, or IT input, making the solution more expensive.
Conversational AI platforms that have no-code/low-code self-serve
capabilities can enable business users to build and deploy voice and
digital bots and context-aware conversational flows in just a few days.

State-of-the-Art Conversational AI

Technology behind conversational bot experiences is based on the latest

advances in artificial intelligence, NLP, sentiment analysis, deep learning, and
intent prediction. Together, these features encourage engagement, improve
customer experience and agent satisfaction, accelerate time to resolution, and

25
grow business value.

Natural Language Processing (NLP)

Most conversational AI uses NLU to intelligently process user inputs

against multiple models, enabling a bot to respond in a more human-like way
to non-transactional journeys. The core technology understands slang, local
nuances, colloquial speech, and can be trained to emulate different tones by
using AI-powered speech synthesis.

Sentiment Analysis

This leading conversational AI technology layer abstracts pre-built

sentiment and social models to prioritize and seamlessly escalate to an agent
when it detects that a customer needs expert advice. Sentiment detection will
recognize, for example, an upset customer and immediately route them to an
agent. You can also prioritize unhappy customers in the system, placing them
in special queues or offering exceptional services.

Deep Learning

This machine learning technique is inspired by the human brain or

‘neural network’ and allows AI to learn by association, just like a child. The
more data AI is exposed to, the better it gets—and the more accurately it can
respond over time. AI models trained with many years of contact center data
from various voice and digital channels result in smarter and more accurate
responses to human inquiries. Response accuracy can be further improved
over time by learning from interactions between customers, chatbots, and
human agents, and optimizing intent models using AI-powered speech
synthesis.

Intent Prediction

Using behavioral analysis and tagging activities, conversational AI

technologies can understand the true meaning behind each consumer’s

26
request. Knowing intent allows companies to deliver the right response at the
right moment through an automated bot or human agent.

The future roadmap for conversational AI platforms includes support

for multiple use cases, multi-domain, and multiple vertical needs, along with
explainable AI.

The Conversational AI Vendor Market

According to Gartner, over 1500 conversational AI providers now offer

various levels of capability, language support, use-cases, and business
models. Sophistication swings widely depending on what’s supported, such
as:

• Number of integrations with back-end systems such as CRM

• Number and type of channels (voice, text-based chatbots, messaging,
etc.).
• Customization of chatbots and virtual assistants for vertical specific use
cases and applications for faster adoption into production
• Number of languages, slang, dialects, local lingo, shorthand, phonetic
spelling, grammatical structures, intents, entity, etc.

Horizontal solutions are the most flexible and controllable but take
longer to implement, while vertical specific ones come with pre-built
capabilities that are a better fit for a specialized use cases in a target domain.
Vendors that offer vertical solutions built on an established horizontal
platform give companies full flexibility in customizing to meet their precise
needs.

How to pick the right Conversational AI Solution?

When it comes to selecting a conversational AI solution, there are a few

key factors to consider.

27
1. First, consider the needs of your business. What questions or tasks do
your customers commonly ask or need help with? What areas of your
business could benefit from automation?
2. Next, evaluate the capabilities of different conversational AI solutions.
Some platforms are better suited for specific tasks or industries, while
others are more versatile.
3. Finally, consider the cost and complexity of implementing different
solutions. Some platforms are more expensive or require more technical
expertise to set up and use.

Once you have a better understanding of your business needs and the
capabilities of different conversational AI solutions, you can begin to narrow
down your options and select the right platform for your business.

1.5 Tune Model Hyperparameters – Azure Machine Learning

Here we will see how to use the Tune Model Hyper parameters component
in Azure Machine Learning designer. The goal is to determine the optimum
hyper parameters for a machine learning model. The component builds and
tests multiple models by using different combinations of settings. It compares
metrics over all models to get the combinations of settings.

The terms parameter and hyper parameter can be confusing. The model's
parameters are what you set in the right pane of the component. Basically,
this component performs a parameter sweep over the specified parameter
settings. It learns an optimal set of hyper parameters, which might be different
for each specific decision tree, dataset, or regression method. The process of
finding the optimal configuration is sometimes called tuning.

The component supports the following method for finding the optimum
settings for a model: integrated train and tune. In this method, you configure
a set of parameters to use. You then let the component iterate over multiple
combinations. The component measures accuracy until it finds a "best"
model. With most learner components, you can choose which parameters

28
should be changed during the training process, and which should remain
fixed.

Depending on how long you want the tuning process to run, you might
decide to exhaustively test all combinations. Or you might shorten the process
by establishing a grid of parameter combinations and testing a randomized
subset of the parameter grid.

How to configure Tune Model Hyper parameters

Learning the optimal hyper parameters for a machine learning model

requires considerable use of pipelines

Train a model by using a parameter sweep

This section describes how to perform a basic parameter sweep, which trains
a model by using the Tune Model Hyper parameters component.

1. Add the Tune Model Hyper parameters component to your pipeline in the
designer.
2. Connect an untrained model to the leftmost input.
3. Add the dataset that you want to use for training, and connect it to the
middle input of Tune Model Hyper parameters.

Optionally, if you have a tagged dataset, you can connect it to the rightmost
input port (Optional validation dataset). This lets you measure accuracy
while training and tuning.

4. In the right panel of Tune Model Hyper parameters, choose a value for
Parameter sweeping mode. This option controls how the parameters are
selected.

• Entire grid: When you select this option, the component loops over a
grid predefined by the system, to try different combinations and identify
the best learner. This option is useful when you don't know what the

29
 best parameter settings might be and want to try all possible
combinations of values.
• Random sweep: When you select this option, the component will
randomly select parameter values over a system-defined range. You
must specify the maximum number of runs that you want the
component to execute. This option is useful when you want to increase
model performance by using the metrics of your choice but still
conserve computing resources.

5. For Label column, open the column selector to choose a single label
column.
6. Choose the number of runs:

• Maximum number of runs on random sweep: If you choose a random

sweep, you can specify how many times the model should be trained,
by using a random combination of parameter values.

7. For Ranking, choose a single metric to use for ranking the models.

When you run a parameter sweep, the component calculates all applicable
metrics for the model type and returns them in the Sweep results report.
The component uses separate metrics for regression and classification
models.

However, the metric that you choose determines how the models are
ranked. Only the top model, as ranked by the chosen metric, is output as
a trained model to use for scoring.

8. For Random seed, enter an integer number as a pseudo random number

generator state used for randomly selecting parameter values over a pre-
defined range. This parameter is only effective if Parameter sweeping
mode is Random sweep.
9. Submit the pipeline.

Results of hyper parameter tuning

30
When training is complete:

• To view the sweep results, you could either right-click the component,
and then select Visualize, or right-click left output port of the
component to visualize.

The Sweep results includes all parameter sweep and accuracy metrics
that apply to the model type, and the metric that you selected for
ranking determines which model is considered "best."

• To save a snapshot of the trained model, select the Outputs+logs tab in

the right panel of the Train model component. Select the Register
dataset icon to save the model as a reusable component.

Train a model by using a parameter sweep

This section describes how to perform a basic parameter sweep, which

trains a model by using the Tune Model Hyper parameters component.

1. Add the Tune Model Hyper parameters component to your pipeline in

the designer.
2. Connect an untrained model to the leftmost input.

Neural Network Regression:

Let us discuss about a component in Azure Machine Learning designer.

This component is used to create regression model using a customizable
neural network algorithm.

Although neural networks are widely known for use in deep learning
and modelling complex problems such as image recognition, they are easily
adapted to regression problems. Any class of statistical models can be termed
a neural network if they use adaptive weights and can approximate non-linear

31
functions of their inputs. Thus, neural network regression is suited to
problems where a more traditional regression model cannot fit a solution.

Neural network regression is a supervised learning method, and therefore

requires a tagged dataset, which includes a label column. Because a
regression model predicts a numerical value, the label column must be a
numerical data type. The model can be trained by providing a model and the
tagged dataset as an input to train the model. The trained model can then be
used to predict values for the new input examples.

Configure Neural Network Regression:

• Create a neural network model using the default architecture: If

you accept the default neural network architecture, use the Properties
pane to set parameters that control the behaviour of the neural
network, such as the number of nodes in the hidden layer, learning
rate, and normalization. Start here if you are new to neural networks.
The component supports many customizations, as well as model
tuning, without deep knowledge of neural networks.
• Define a custom architecture for a neural network: Use this option
if you want to add extra hidden layers, or fully customize the network
architecture, its connections, and activation functions. This option is
best if you are already somewhat familiar with neural networks. You
use the Net# language to define the network architecture.

1. Create a neural network model using the default architecture:

• Add the Neural Network Regression component to your pipeline in the

designer. You can find this component under Machine Learning,
Initialize, in the Regression category.
• Indicate how you want the model to be trained, by setting the Create
trainer mode option.
o Single Parameter: Choose this option if you already know how you
want to configure the model.
o Parameter Range: Select this option if you are not sure of the best

32
 parameters, and want to run a parameter sweep. Select a range of
values to iterate over, and the Tune Model Hyper parameters iterates
over all possible combinations of the settings you provided to
determine the hyper parameters that produce the optimal results.
• In Hidden layer specification, select fully connected case. This option
creates a model using the default neural network architecture, which
for a neural network regression model, has these attributes:
o The network has exactly one hidden layer.
o The output layer is fully connected to the hidden layer and the
hidden layer is fully connected to the input layer.
o The number of nodes in the hidden layer can beset by the user
(default value is 100).

Because the number of nodes in the input layer is determined by the

number of features in the training data, in a regression model there
can be only one node in the output layer.

• For Number of hidden nodes, type the number of hidden nodes. The
default is one hidden layer with 100 nodes. (This option is not available
if you define a custom architecture using Net#.)
• For learning rate, type a value that defines the step taken a teach
iteration, before correction. A larger value for learning rate can cause
the model to converge faster, but it can overshoot local minima.
• For Number of learning iterations, specify the maximum number of
times the algorithm processes the training cases.
• For The momentum, type a value to apply during learning as a weight
on nodes from previous iterations.
• Select the option, Shuffle examples, to change the order of cases
between iterations. If you deselect this option, cases are processed in
exactly the same order each time you run the pipeline.
• For Random number seed, you can optionally type a value to use as the
seed. Specifying a seed value is useful when you want to ensure
repeatability across runs of the same pipeline.
• Connect training data set and train the model:

33
 o If you set Create trainer mode to Single Parameter, connect a tagged
data set and the Train Model component.
o If you set Create trainer mode to Parameter Range, connect a tagged
data set and train the model by using Tune Model Hyper parameters.
• Submit the Pipeline

Course 2

Data Analyst Associate (DA -100)

Introduction to Microsoft data analytics

Before data can be used to tell a story, it must be run through a process
that makes it usable in the story. Data analysis is the process of identifying,
cleaning, transforming, and modelling data to discover meaningful and useful

34
information. The data is then crafted into a story through reports for analysis
to support the critical decision-making process.

As the world becomes more data-driven, storytelling through data

analysis is becoming a vital component and aspect of large and small
businesses. It is the reason that organizations continue to hire data analysts.

Data-driven businesses make decisions based on the story that their

data tells, and in today's data-driven world, data is not being used to its full
potential, a challenge that most businesses face. Data analysis is, and should
be, a critical aspect of all organizations to help determine the impact to their
business, including evaluating customer sentiment, performing market and
product research, and identifying trends or other data insights.

While the process of data analysis focuses on the tasks of cleaning,

modelling, and visualizing data, the concept of data analysis and its
importance to business should not be understated. To analyse data, core
components of analytics are divided into the following categories:

• Descriptive
• Diagnostic
• Predictive
• Prescriptive
• Cognitive
• Descriptive analytics: Descriptive analytics help answer questions about
what has happened based on historical data. Descriptive analytics
techniques summarize large datasets to describe outcomes to
stakeholders. By developing key performance indicators (KPIs), these
strategies can help track the success or failure of key objectives. Metrics
such as return on investment (ROI) are used in many industries, and
specialized metrics are developed to track performance in specific
industries.An example of descriptive analytics is generating reports to
provide a view of an organization's sales and financial data.
• Diagnostic analytics: Diagnostic analytics help answer questions about
why events happened. Diagnostic analytics techniques supplement basic

35
 descriptive analytics, and they use the findings from descriptive analytics
to discover the cause of these events. Then, performance indicators are
further investigated to discover why these events improved or became
worse. Generally, this process occurs in three steps:
1. Identify anomalies in the data. These anomalies might be unexpected
changes in a metric or a particular market.
2. Collect data that's related to these anomalies.
3. Use statistical techniques to discover relationships and trends that
explain these anomalies.
• Predictive analytics: Predictive analytics help answer questions about
what will happen in the future. Predictive analytics techniques use
historical data to identify trends and determine if they're likely to recur.
Predictive analytical tools provide valuable insight into what might happen
in the future. Techniques include a variety of statistical and machine
learning techniques such as neural networks, decision trees, and
regression.
• Prescriptive analytics: Prescriptive analytics help answer questions
about which actions should be taken to achieve a goal or target. By using
insights from prescriptive analytics, organizations can make data-driven
decisions. This technique allows businesses to make informed decisions in
the face of uncertainty. Prescriptive analytics techniques rely on machine
learning as one of the strategies to find patterns in large datasets. By
analysing past decisions and events, organizations can estimate the
likelihood of different outcomes.
• Cognitive analytics: Cognitive analytics attempt to draw inferences from
existing data and patterns, derive conclusions based on existing knowledge
bases, and then add these findings back into the knowledge base for future
inferences, a self-learning feedback loop. Cognitive analytics help you learn
what might happen if circumstances change and determine how you might
handle these situations. Inferences aren't structured queries based on a
rules database; rather, they're unstructured hypotheses that are gathered
from several sources and expressed with varying degrees of confidence.
Effective cognitive analytics depend on machine learning algorithms, and
will use several natural language processing concepts to make sense of

36
 previously untapped data sources, such as call centre conversation logs
and product reviews.

2. Prepare data for analysis

Your manager wants to see a report on your latest sales and profit
figures by the end of the day. But the latest data is in files on your laptop. In
the past, it’s taken hours to create a report, and you’re beginning to feel
anxious. With Power BI, you can create a stunning report and share it in
Microsoft Teams in no time!

Here, we upload an Excel file, create a new report, and share it with
colleagues in Microsoft Teams, all from within Power BI. You'll learn how to:

• Prepare your data in Excel.

• Download sample data.
• Build a report in the Power BI service.
• Pin the report visuals to a dashboard.
• Share a link to the dashboard.
• Share the dashboard in Microsoft Teams

Prerequisites

37
 • Sign up for the Power BI service.
• Download the Financial Sample workbook and save it your computer or
to OneDrive for Business.

PRACTICAL

Prepare data in Excel

Let’s take a simple Excel file as an example.

1. Before you can load your Excel file into Power BI, you must organize your
data in a flat table. In a flat table, each column contains the same data
type; for example, text, date, number, or currency. Your table should have
a header row, but not any columns or rows that display totals.

1. Next, format your data as a table. In Excel, on the Home tab, in the
Styles group, select Format as Table.
2. Select a table style to apply to your worksheet. Your Excel worksheet is
now ready to load into Power BI.

38
Upload your Excel file to the Power BI service

The Power BI service connects to many data sources, including Excel files that
live on your computer.

3. To get started, sign in to the Power BI service. If you haven’t signed up,
you can do so for free.

2. In My workspace, select New > Upload a file.

1. Select Local File, browse to where you saved the Financial Sample
Excel file, and select Open.
2. On the Local File page, select Import: Now you have a Financial

39
 Sample dataset. Power BI also automatically created a blank
dashboard. If you don't see the dashboard, refresh your browser.

3. You want to create a report. Still in My workspace, select New >

Report.

4. In the Select a dataset to create a report dialog box, select your

Financial Sample dataset > Create.

40
Build your report: The report opens in editing view and displays the blank
report canvas. On the right are the Visualizations, Filters, and Fields panes.
Your Excel workbook table data appears in the Fields pane. At the top is the
name of the table, financials. Under that, Power BI lists the column headings
as individual fields. You see the Sigma symbols in the Fields list? Power BI
has detected that those fields are numeric. Power BI also indicates a
geographic field with a globe symbol.

To have more room for the report canvas, select Hide the navigation
pane, and minimize the Filters pane.

41
 Now you can begin to create visualizations. Let's say your manager
wants to see profit over time. In the Fields pane, drag Profit to the report
canvas. By default, Power BI displays a column chart with one column.

42
 • Drag Date to the report canvas.
• Power BI updates the column chart to show profit by date.

December 2014 was the most profitable month.

4. Model data in Power BI: Often, you'll connect to multiple data sources
to create your reports. All that data needs to work together to create a
cohesive report. Modelling is how to get your connected data ready for
use.

Tasks in this module:

• Create relationships between your data sources

• Create a new field with calculated columns

43
• Optimize data by hiding fields and sorting visualization data
• Create a measure to perform calculations on your data
• Use a calculated table to create a relationship between two tables
• Format time-based data so that you can drill down for more details

Fig: Create relationships between your data sources

Fig: Create a new field with calculated columns

44
 Fig: Optimize data by hiding fields and sorting visualization data

Fig: Create a measure to perform calculations on your data

Fig: Use a calculated table to create a relationship between two tables

45
 Fig: Format time-based data so that you can drill down for more details

In Power BI, you can create a relationship to create a logical connection

between different data sources. A relationship enables Power BI to connect
tables to one another so that you can create visuals and reports. This module
describes data-centric relationships and how to create relationships when
none exists.

5. Visualize data in Power BI

Visualizations are used to effectively present your data and are the
basic building blocks of any Business Intelligence tool. Power BI
contains various default data visualization components that include
simple bar charts to pie charts to maps, and also complex models such
as waterfalls, funnels, gauges, and many other components.

46
 In Power BI, you can create visualization in two ways. First is by adding
from the right-side pane to Report Canvas. By default, it is the table type
visualization, which is selected in Power BI. Another way is to drag the fields
from right side bar to the axis and value axis under Visualization. You can
add multiple fields to each axis as per the requirement.

In Power BI, it is also possible to move your visualization on the

reporting canvas by clicking and then dragging it. You can also switch
between different type of charts and visualizations from the Visualization
pane. Power BI attempts to convert your selected fields to the new visual type
as closely as possible.

Creating Map Visualizations

In Power BI, we have two types of map visualization - bubble maps and
shape maps. If you want to create a bubble map, select the map option from
the visualization pane.

47
 To use a bubble map, drag the map from Visualizations to the Report
Canvas. To display values, you have to add any location object to the axis.

In the value fields, you can see that it accepts values axis such as City
and State and or you can also add longitude and latitude values. To change
the bubble size, you need to add a field to the value axis.

You can also use a filled map in data visualization, just by dragging the
filled map to the Report Canvas.

48
Note − If you see a warning symbol on top of your map visualization, it means
that you need to add more locations to your map chart.

Using Combination Charts

In data visualization, it is also required to plot multiple measures in a

single chart. Power BI supports various combination chart types to plot
measure values. Let us say you want to plot revenue and unit sold in one
chart. Combination charts are the most suitable option for these kinds of
requirement.

One of the most common Combination charts in Power BI is Line and

Stacked column charts. Let us say we have a revenue field and we have added
a new data source that contains customer-wise unit quantity and we want to
plot this in our visualization.

49
 Once you add a data source, it will be added to the list of fields on the
right side. You can add units to the column axis as shown in the following
screenshot.

You have other type of combine chart that you can use in Power BI -
Line and Clustered Column.

Using Tables

In Power BI, when you add a dataset to your visualization, it adds a

table chart to the Report canvas. You can drag the fields that you want to add
to the report. You can also select the checkbox in front of each field to add
those to the Report area.

50
 With the numerical values in a table, you can see a sum of values at
the bottom.

You can also perform a sort in the table using an arrow key at the top
of the column. To perform ascending/descending sort, just click the arrow
mark, and the values in the column will be sorted.

The order of the columns in a table is determined by the order in the

value bucket on the right side. If you want to change the order, you can delete
any column and add the other one.

51
 You can also undo summarize or apply different aggregate function on
numerical values in the table. To change the aggregation type, click the arrow
in the value bucket in front of the measure and you will see a list of formulas
that can be used

52
 Another table type in Power BI is the matrix table that provides a lot of
features such as auto sizing, column tables, and setting colors, etc.

Modify Colours in Charts

In Power BI, you can also modify the colours in the chart. When you
select any visualization, it has an option to change the colour. Following
options are available under the Format tab −

• Legend
• Data colours
• Detail Label
• Title
• Background
• Lock Aspect
• Border
• General

To open these options, go to the Format tab as shown in the following

screenshot. Once you click, you can see all the options available.

53
When you expand the Legend field, you have an option where you want to
display the legend. You can select −

• Position
• Title
• Legend Name
• Colour
• Text Size
• Font Family

54
 Similarly, you have data colours. In case, you want to change the colour
of any data field, you can use this option. It shows all objects and their
corresponding colours in the chart.

You also have Analytics feature in the tool, where you can draw lines
as per requirement in data visualization. You have the following line types in
data visualization −

• Constant Line
• Min Line
• Max Line
• Average Line
• Median Line
• Percentile Line

55
 You can opt for a dashed, dotted, or a solid line. You can select
Transparency level, color, and position of the line. You can also switch on/off
data label for this line.

Adding Shapes, Images and Text box

Sometimes it is required that you need to add static text, images, or

shapes to your visualization. In case you want to add header/footer or any
static signatures, messages to data visualization this option can be used.

You can also add URLs in the text box and Power BI uses those link to
make it live.

To add shapes, images and text box, navigate to the Home tab and at
the top you will find an option to add images.

You can insert different shapes in data visualization. To see the available
shapes, click the arrow next to the Shapes button.

56
 When you click on the text box, it adds a text box in your Report canvas.
You can enter any text in the text box and use the rich text editor to make
formatting changes.

Similarly, images can be added to data visualization to add logos or

other images to data visualization. When you click the Image option, it asks
for a path to pass the image file.

You can add shapes by selecting any shape from the dropdown list. You
can also resize it using different options.

57
Styling Reports

In Power BI, you have flexible options to adjust the page layout and
formatting such as orientation and page size of your report. Navigate to Page
View menu from the Home tab and the following options are provided.

• Fit to Page
• Fit to Width
• Actual Size

58
 By default, the page size in a report is 16:9; however, it is also possible
to change the page size of the report. To change the page size, navigate to the
Visualization pane and select Paint brush.

Note − To change page size, no visualization should be added to the Report

canvas. You have the following options available under Page layout −

• Page Information
• Page Size
• Page Background

Under Page Information, you have Name and Q&A.

Under Page Size, you can select from the following options −

• Type
• Width
• Height

Under Page Background, you can select from the following options:

59
 • Colour
• Transparency
• Add Image

Duplicating Reports

In some scenarios, you may want to use the same layout and visuals for
different pages. Power BI provides an option to create a copy of the page. When
you use Duplicate Page option, a new page is added with similar layout and
visuals.

To duplicate a page, right-click the Page and select Duplicate Page option.
This will create a copy of the same page with the name - Duplicate of Page1.

5.Data Analysis in Power BI

60
 Microsoft Power BI is the leading data analytics, business intelligence
and reporting tool in Gartner's latest Magic Quadrant. Beyond the opinion of
a leading global technology consulting company, Power BI is also one of the
most widely used data analytics tools by professionals because of its business
approach.

The tool enables data scientists and analysts to turn data sets into
engaging, interactive and insight-enhancing reports, dashboards and visuals
used to monitor the course of business strategies. Power BI allows us to
develop customized solutions for each of our customers.

Data analytics with Power BI promotes the discovery of information and

the democratisation of data through its unique data visualisation capabilities,
that improve the quality of companies' informative systems and help
executives make data-driven decisions.

Power BI streamlines data analysts' work and makes it easy to connect,

transform and visualise data.

Nevertheless, there are many other similar tools on the market. For example,
Excel also by Microsoft is another essential tool for data experts. In fact, using
Excel and Power BI together is very common.

Important reasons why you should do data analytics with Power BI

1. Data connection

One of Power BI's greatest advantages is its extensive data connectivity.

The tool connects to multiple tabular databases and integrates with a host
of corporate tools and systems to make importing and exporting data,
dashboards and reports as simple and fast as possible.

2. Data visualization

Power BI is one of the most complete platforms for data visualisation.

61
 In the tool's appsource you will find a lot of Power BI visuals validated by
Microsoft, but you can also create your own custom visuals.

In addition, Power BI adds new visuals from time to time and you can even
extend its visualisation capabilities with Zebra BI, which works both for
Excel and Power BI.

3. Data visualization

Power BI is one of the most complete platforms for data visualisation.

In the tool's appsource you will find a lot of Power BI visuals validated by
Microsoft, but you can also create your own custom visuals.

In addition, Power BI adds new visuals from time to time and you can even
extend its visualisation capabilities with Zebra BI, which works both for
Excel and Power BI.

4. Advanced analytics

Power BI is the optimal platform for increasing the value of your regular
Excel data analysis with advanced analytics. You can enrich business
data by ingesting, transforming and integrating data services to other
Microsoft suite's tools.

5. Data governance

For anyone who works with data, data governance is a must to ensure
the smooth running of any type of process, especially in the business
environment, since organisations often have a large amount of data that,
when not well organised, can lose all its value.

Power BI includes features that support data control, authority and

management. However, the tool has limited data governance capabilities
and some organisations require specialised data governance solutions to
work with Power BI.

Find out how you can apply data governance policies to Power BI

62
 With Power BI Viewer you can organise your reports by category, assign
permissions according to users' role, access all your reports without having
a Power BI licence and much more.

With Power BI Analytics you will be able to analyse the activity of all your
users with a historical repository with no space or time limits.

6. Data exploration

Power BI contains extensive data exploration options as well as automated

queries. With this tool, discovering insights from data will be much easier.
It is also the ideal platform for working with a top-down methodology.

7. UX & UI

Finally, it should be noted that Power BI was conceived as an enterprise

tool designed to be used by business users. Clearly, its main users are data
analysts and BI consultants. However, the business conception of the
platform makes it one of the BI tools with the best usability and user
interface. In addition, with Power BI you can adapt your reports to your
brand image and automate the process by designing themes that can be
applied to all your reports.

8. Manage workspaces and datasets in Power BI

Workspaces are places to collaborate with colleagues to create collections

of dashboards, reports, datasets, and paginated reports. Here we describe
how to manage, access and how to use them to create and distribute apps.

Creating a workspace

63
Working with workspaces

Here are some useful tips about working with workspaces.

• Use granular workspace roles for flexible permissions management in

the workspaces: Admin, Member, Contributor, and Viewer.
• Contact list: Specify who receives notification about workspace
activity.
• Create template apps: You can create template apps in workspaces.
Template apps are apps that you can distribute to customers outside of
your organization. Those customers can then connect to their own data
with your template app.
• Share datasets: You can share datasets between workspaces.

Workspace contact list

The Contact list feature allows you to specify which users receive notification
about issues occurring in the workspace. By default, any user or group
specified as a workspace admin in the workspace is notified. You can add to
that list. Users or groups in the contact list are also listed in the user interface

64
(UI) of the workspaces, so workspace end-users know whom to contact.

Create a workspace in Power BI

Here, we discuss how to create workspaces, spaces to collaborate with

colleagues. In them, you create collections of dashboards, reports, and
paginated reports. If you want, you can also bundle that collection into an
app and distribute it to a broader audience.

Create a workspace

1. Select Workspaces > Create workspace.

65
2. Give the workspace a unique name. If the name isn't available, edit it to
come up with a name that's unique.

When you create an app from the workspace, by default it will have the
same name and icon as the workspace. You can change both when you
create the app.

3. Here are some optional settings for your workspace.

• Upload a Workspace image. Files can be .png or .jpg format. File size
has to be less than 45 KB.
• Specify a Workspace One Drive to use a Microsoft 365 Group file storage
location (provided by SharePoint).
• Add a Contact list, the names of people to contact for information about
the workspace. By default, the workspace admins are the contacts.
• Allow contributors to update the app for the workspace
• Assign the workspace to a Premium capacity.
• Connect the workspace to an Azure Data Lake Gen2 storage account
(in preview).

4. Select Save.

Power BI creates the workspace and opens it. You see it in the list of
workspaces you’re a member of. Microsoft 365 and OneDrive: Power BI
doesn't create a Microsoft 365 group behind the scenes when you create a

66
 workspace. All workspace administration is in Power BI. Still, you might
find it useful to have a OneDrive associated with the workspace.

• You can manage user access to content through Microsoft 365 groups,
if you want. You add a Microsoft 365 group in the workspace access
list.

Power BI doesn't synchronize between Microsoft 365 group membership

and permissions for users or groups with access to the workspace. You
can synchronize them: Manage workspace access through the same
Microsoft 365 group whose file storage you configure in this setting.

• You can also store Power BI content in OneDrive for Business. With
the Workspace One Drive feature in workspaces, you can configure a
Microsoft 365 group whose SharePoint Document Library file storage is
available to workspace users. You create the group outside of Power BI.

Note :

Power BI lists all Microsoft 365 groups that you're a member of in the
workspaces list.

Roles and licenses

Roles let you manage who can do what in workspaces, so team members can
collaborate. To grant access to a workspace, assign those user groups or
individuals to one of the workspace roles: Admin, Member, Contributor, or
Viewer.

• Licensing enforcement: Publishing reports to a workspace enforces

existing licensing rules. Users collaborating in workspaces or sharing
content to others in the Power BI service need a Power BI Pro or Premium
Per User (PPU) license. Users without a Pro or PPU license see the error
"Only users with Power BI Pro licenses can publish to this workspace."
• Read-only workspaces: The Viewer role in workspaces gives users read-
only access to the content in a workspace.

67
• Users without a Pro or Premium Per User (PPU) license can access a
workspace if the workspace is in a Power BI Premium capacity, but only if
they have the Viewer role.
• Allow users to export data: Even users with the Viewer role in the
workspace can export data if they have build permission on the datasets
in that workspace.
• Assign user groups to workspace roles: You can add Active Directory
security groups, distribution lists, or Microsoft 365 groups to these roles,
for easier user management.

Administering and auditing workspaces

Administration for workspaces is in the Power BI admin portal. Power

BI admins decide who in an organization can create workspaces and
distribute apps.

Admins can also see the state of all the workspaces in their
organization. They can manage, recover, and even delete workspaces.

Auditing

Power BI audits the following activities for workspaces.

Considerations and limitations

Limitations to be aware of:

• Workspaces can contain a maximum of 1,000 datasets, or 1,000 reports

per dataset.

68
 • Power BI publisher for Excel isn't supported.
• Certain special characters aren't supported in workspace names when
using an XMLA endpoint. As a workaround, use URL encoding of special
characters, for example, for a forward slash /, use %2F.

5. Key Influencers Visualizations Tutorial - Power BI

The key influencers visual helps you understand the factors that drive a
metric you're interested in. It analyses your data, ranks the factors that
matter, and displays them as key influencers. For example, suppose you
want to figure out what influence’s employee turnover, which is also known
as churn. One factor might be employment contract length, and another
factor might be commute time.

When to use key influencers

The key influencers visual is a great choice if you want to:

• See which factors affect the metric being analysed.

• Contrast the relative importance of these factors. For example, do short-
term contracts affect churn more than long-term contracts?

Features of the key influencers visual

69
1. Tabs: Select a tab to switch between views. Key influencers show you the
top contributors to the selected metric value. Top segments show you the
top segments that contribute to the selected metric value. A segment is
made up of a combination of values. For example, one segment might be
consumers who have been customers for at least 20 years and live in the
west region.
2. Drop-down box: The value of the metric under investigation. In this
example, look at the metric Rating. The selected value is Low.
3. Restatement: It helps you interpret the visual in the left pane.
4. Left pane: The left pane contains one visual. In this case, the left pane
shows a list of the top key influencers.
5. Restatement: It helps you interpret the visual in the right pane.
6. Right pane: The right pane contains one visual. In this case, the column
chart displays all the values for the key influencer Theme that was
selected in the left pane. The specific value of usability from the left pane
is shown in green. All the other values for Theme are shown in black.
7. Average line: The average is calculated for all possible values for Theme
except usability (which is the selected influencer). So, the calculation

70
 applies to all the values in black. It tells you what percentage of the other
Themes had a low rating. In this case 11.35% had a low rating (shown by
the dotted line).
8. Check box: Filters out the visual in the right pane to only show values that
are influencers for that field. In this example, the visual is filtered to
display usability, security, and navigation.

Analyse a metric that is categorical

Here, we learn how to create a key influencer visual with a categorical

metric. Then follow the steps to create one.

1. Your Product Manager wants you to figure out which factors lead
customers to leave negative reviews about your cloud service. To follow
along in Power BI Desktop, open the Customer Feedback PBIX file
2. Under Build visual on the Visualizations pane, select the Key
influencers icon.
3. Move the metric you want to investigate into the Analyse field. To see what
drives a customer rating of the service to be low, select Customer Table >
Rating.
4. Move fields that you think might influence Rating into the Explain by
field. You can move as many fields as you want. In this case, start with:

• Country-Region
• Role in Org
• Subscription Type
• Company Size
• Theme

5. Leave the Expand by field empty. This field is only used when analysing a
measure or summarized field.
6. To focus on the negative ratings, select Low in the What influences Rating
to be drop-down box.

71
 The analysis runs on the table level of the field that's being analysed.
In this case, it's the Rating metric. This metric is defined at a customer level.

72
Each customer has given either a high score or a low score. All the explanatory
factors must be defined at the customer level for the visual to make use of
them.

In this case, each customer assigned a single theme to their rating.

Similarly, customers come from one country, have one membership type, and
hold one role in their organization. The explanatory factors are already
attributes of a customer, and no transformations are needed. The visual can
make immediate use of them.

Interpret categorical key influencers

Let's take a look at the key influencers for low ratings.

Top single factor that influences the likelihood of a low rating

The customer in this example can have three roles: consumer, administrator,
and publisher. Being a consumer is the top factor that contributes to a low
rating.

More precisely, your consumers are 2.57 times more likely to give your service
a negative score. The key influencers chart lists Role in Org is consumer first

73
in the list on the left. By selecting Role in Org is consumer, Power BI shows
more details in the right pane. The comparative effect of each role on the
likelihood of a low rating is shown.

• 14.93% of consumers give a low score.

• On average, all other roles give a low score 5.78% of the time.
• Consumers are 2.57 times more likely to give a low score compared to all
other roles. You can determine this score by dividing the green bar by
the red dotted line.

Second single factor that influences the likelihood of a low rating

The key influencers visual compares and ranks factors from many
different variables. The second influencer has nothing to do with Role in Org.
Select the second influencer in the list, which is Theme is usability.

The second most important factor is related to the theme of the

74
customer’s review. Customers who commented about the usability of the
product were 2.55 times more likely to give a low score compared to customers
who commented on other themes, such as reliability, design, or speed.

Between the visuals, the average, which is shown by the red dotted line,
changed from 5.78% to 11.35%. The average is dynamic because it's based
on the average of all other values. For the first influencer, the average
excluded the customer role. For the second influencer, it excluded the
usability theme.

Select the Only show values that are influencers check box to filter
by using only the influential values. In this case, they're the roles that drive a
low score. 12 themes are reduced to the four that Power BI identified as the
themes that drive low ratings.

Interact with other visuals

Every time you select a slicer, filter, or other visual on the canvas, the

75
key influencers visual reruns its analysis on the new portion of data. For
example, you can move Company Size into the report and use it as a slicer.
Use it to see if the key influencers for your enterprise customers are different
than the general population. An enterprise company size is larger than 50,000
employees.

Select >50,000 to rerun the analysis, and you can see that the influencers
changed. For large enterprise customers, the top influencer for low ratings
has a theme related to security. You might want to investigate further to see
if there are specific security features your large customers are unhappy about.

Interpret continuous key influencers

So far, you've seen how to use the visual to explore how different
categorical fields influence low ratings. It's also possible to have continuous

76
factors such as age, height, and price in the Explain by field. Let’s look at
what happens when Tenure is moved from the customer table into Explain
by. Tenure depicts how long a customer has used the service.

As tenure increases, the likelihood of receiving a lower rating also

increases. This trend suggests that the longer-term customers are more likely
to give a negative score. This insight is interesting, and one that you might
want to follow up on later.

The visualization shows that every time tenure goes up by 13.44

months, on average the likelihood of a low rating increases by 1.23 times. In
this case, 13.44 months depict the standard deviation of tenure. So, the
insight you receive looks at how increasing tenure by a standard amount,
which is the standard deviation of tenure, affects the likelihood of receiving a
low rating.

The scatter plot in the right pane plots the average percentage of low
ratings for each value of tenure. It highlights the slope with a trend line.

Binned continuous key influencers

In some cases, you may find that your continuous factors were
automatically turned into categorical ones. If the relationship between the

77
variables isn't linear, we can't describe the relationship as simply increasing
or decreasing.

We run correlation tests to determine how linear the influencer is with

regard to the target. If the target is continuous, we run Pearson correlation
and if the target is categorical, we run Point Biserial correlation tests. If we
detect the relationship isn't sufficiently linear, we conduct supervised binning
and generate a maximum of five bins. To figure out which bins make the most
sense, we use a supervised binning method that looks at the relationship
between the explanatory factor and the target being analysed.

Interpret measures and aggregates as key influencers

You can use measures and aggregates as explanatory factors inside

your analysis. For example, you might want to see what effect the count of
customer support tickets or the average duration of an open ticket has on the
score you receive.

In this case, you want to see if the number of support tickets that a
customer has influences the score they give. Now you bring in Support Ticket
ID from the support ticket table. Because a customer can have multiple
support tickets, you aggregate the ID to the customer level. Aggregation is
important because the analysis runs on the customer level, so all drivers must
be defined at that level of granularity.

Let's look at the count of IDs. Each customer row has a count of support
tickets associated with it. In this case, as the count of support tickets
increases, the likelihood of the rating being low goes up 4.08 times. The visual
on the right shows the average number of support tickets by different Rating
values evaluated at the customer level.

78
Interpret the results: Top segments

You can use the Key influencers tab to assess each factor individually.
You also can use the Top segments tab to see how a combination of factors
affects the metric that you're analysing.

Top segments initially show an overview of all the segments that Power
BI discovered. The following example shows that six segments were found.
These segments are ranked by the percentage of low ratings within the
segment. Segment 1, for example, has 74.3% customer ratings that are low.
The higher the bubble, the higher the proportion of low ratings. The size of
the bubble represents how many customers are within the segment.

79
 Selecting a bubble displays the details of that segment. If you select
Segment 1, for example, you find that it's made up of relatively established
customers. They've been customers for over 29 months and have more than
four support tickets. Finally, they're not publishers, so they're either
consumers or administrators.

In this group, 74.3% of the customers gave a low rating. The average
customer gave a low rating 11.7% of the time, so this segment has a larger
proportion of low ratings. It's 63 percentage points higher. Segment 1 also
contains approximately 2.2% of the data, so it represents an addressable
portion of the population.

Adding counts

Sometimes an influencer can have a significant effect but represent

little of the data. For example, Theme is usability is the third biggest
influencer for low ratings. However, there might have only been a handful of
customers who complained about usability. Counts can help you prioritize
which influencers you want to focus on.

80
You can turn on counts through the Analysis card of the formatting pane.

81
 After counts are enabled, you’ll see a ring around each influencer’s
bubble, which represents the approximate percentage of data that influencer
contains. The more of the bubble the ring circles, the more data it contains.
We can see that Theme is usability contains a small proportion of data.

You can also use the Sort by toggle in the bottom left of the visual to
sort the bubbles by count first instead of impact. Subscription Type is
Premier is the top influencer based on count.

82
Having a full ring around the circle means the influencer contains 100% of
the data. You can change the count type to be relative to the maximum
influencer using the Count type dropdown in the Analysis card of the
formatting pane. Now the influencer with the most amount of data will be
represented by a full ring and all other counts will be relative to it.

Analyse a metric that is numeric

83
 If you move an unsummarized numerical field into the Analyse field,
you have a choice how to handle that scenario. You can change the behaviour
of the visual by going into the Formatting Pane and switching between
Categorical Analysis Type and Continuous Analysis Type.

A Categorical Analysis Type behaves as described above. For instance,

if you were looking at survey scores ranging from 1 to 10, you could ask ‘What
influences Survey Scores to be 1?’

A Continuous Analysis Type changes the question to a continuous

one. In the example above, our new question would be ‘What influences
Survey Scores to increase/decrease?’

This distinction is helpful when you have lots of unique values in the
field, you're analysing. In the example below, we look at house prices. It isn't
meaningful to ask ‘What influences House Price to be 156,214?’ as that is very
specific and we're likely not to have enough data to infer a pattern.

Instead we may want to ask, ‘What influences House Price to increase’?

which allows us to treat house prices as a range rather than distinct values.

Interpret the results: Key influencers

In this scenario, we look at ‘What influences House Price to increase’. A

number of explanatory factors could impact a house price like Year Built
(year the house was built), Kitchen Qual (kitchen quality), and
YeaRemodAdd (year the house was remodelled).

84
In the example below, we look at our top influencer which is kitchen quality
being Excellent. The results are similar to the ones we saw when we were
analysing categorical metrics with a few important differences:

• The column chart on the right is looking at the averages rather than
percentages. It therefore shows us what the average house price of a
house with an excellent kitchen is (green bar) compared to the average
house price of a house without an excellent kitchen (dotted line)
• The number in the bubble is still the difference between the red dotted
line and green bar but it’s expressed as a number ($158.49K) rather than
likelihood (1.93x). So, on average, houses with excellent kitchens are
almost $160K more expensive than houses without excellent kitchens.

In the example below, we look at the impact a continuous factor (year house
was remodelled) has on house price. The differences compared to how we
analyse continuous influencers for categorical metrics are as follows:

• The scatter plot in the right pane plots the average house price for each
distinct value of year remodelled.
• The value in the bubble shows by how much the average house price

85
 increases (in this case $2.87k) when the year the house was remodelled
increases by its standard deviation (in this case 20 years)

Finally, in the case of measures, we're looking at the average year a house
was built. The analysis is as follows:

• The scatter plot in the right pane plots the average house price for each
distinct value in the table
• The value in the bubble shows by how much the average house price
increases (in this case $1.35K) when the average year increases by its
standard deviation (in this case 30 years)

86
Interpret the results: Top Segments

Top segments for numerical targets show groups where the house
prices on average are higher than in the overall dataset. For example, below
we can see that Segment 1 is made up of houses where Garage Cars (number
of cars the garage can fit) is greater than 2 and the Roof Style is Hip. Houses
with those characteristics have an average price of $355K compared to the
overall average in the data which is $180K.

87
8 Smart Narratives Tutorial - Power BI | Microsoft Docs

The smart narrative visualization helps you quickly summarize visuals and
reports. It provides relevant innovative insights that you can customize.

Use smart narrative summaries in your reports to address key takeaways, to

point out trends, and to edit the language and format for a specific audience.
In PowerPoint, instead of pasting a screenshot of your report's key takeaways,
you can add narratives that are updated with every refresh. You can use the
summaries to understand the data, get to key points faster, and explain the
data to others.

88
Get started

How to use smart narratives is described below.

Choose the Products tab, then select the Smart narrative icon in the
Visualizations pane to automatically generate a summary.

smart narratives can automatically generate a summary of the report's visuals

that address revenue, website visits, and sales. To generate a smart narrative
of a visualization, right-click it and then select Summarize. The smart
narrative also shows the expected range of values for these metrics.

Edit the summary

The smart narrative summary is highly customizable. You can edit or add to
the existing text by using the text box commands. For example, you can make
the text bold or change its colour.

To customize the summary or add your own insights, use dynamic values.

89
You can map text to existing fields and measures or use natural language to
define a new measure to map to text.

To format a dynamic value, select the value in the summary to see your editing
options on the Review tab. Or in the text box, next to the value that you want
to edit, select the edit button.

You can also use the Review tab to review, delete, or reuse previously
defined values. Select the plus sign (+) to insert the value into the summary.
You can also show automatically generated values by turning on the option
at the bottom of the Review tab.

Sometimes a hidden-summary symbol appears in the smart narrative.

It indicates that current data and filters produce no result for the value. A
summary is empty when no insights are available. Hidden-summary symbols
are visible only when you try to edit a summary.

Visual interactions

A summary is dynamic. It automatically updates the generated text and

dynamic values when you cross-filter. For example, if you select electronics
products in the sample file's donut chart, the rest of the report is cross-
filtered, and the summary is also cross-filtered to focus on the electronics
products.

90
 You can also do more advanced filtering. For example, in the sample
file, look at the visual of trends for multiple products. If you're interested only
in a trend for a certain quarter, then select the relevant data points to update
the summary for that trend.

There's a limit to the number of summaries that can be generated so

91
Smart Narratives picks the most interesting things to summarize about the
visual. Smart Narratives generates up to four summaries per visual and up to
16 per page. The summaries that are generated for a page depend on the
location and size of visuals and it avoids generating the same kind of
summaries for different visuals. Therefore summarizing just the visual can
generate more summaries that aren't present while summarizing the whole
page.

Considerations and limitations

The smart narrative feature doesn't support the following functionality:

• Pinning to a dashboard
• Using dynamic values and conditional formatting (for example, data
bound title)
• Publish to Web
• Power BI Report Server
• On-premises Analysis Services
• Live Connection to Azure Analysis Services or SQL Server Analysis
Services
• Multidimensional Analysis Services data sources
• Key influencers visual with a categorical metric or unsummarized
numerical field as 'Analyse' field from a table:
o that contains more than one primary key
o without a primary key, and measures or aggregates as 'Explain by'
fields
• Map visual with non-aggregated latitude or longitude
• Multi-row card with more than three categorical fields
• Cards with non-numeric measures
• Tables, matrices, R visuals or Python visuals, custom visuals
• Summaries of visuals whose columns are grouped by other columns and
for visuals that are built on a data group field
• Cross-filtering out of a visual

92
 • Renaming dynamic values or editing automatically generated dynamic
values
• Summaries of visuals that contain on-the-fly calculations like QnA
arithmetic, complex measures such as percentage of grand total and
measures from extension schemas.

PRACTICAL

1. Describe Artificial intelligence workloads and considerations

Artificial Intelligence (AI) is computers thinking and acting in a way that

simulates a human. AI is a technology that takes information from its
environment and responds based on what it learns. The goal of AI is to
create a machine that can mimic human behaviour.

AI is more than learning—it is knowledge representation, reasoning, and

abstract thinking. Machine learning (ML) is the subset of AI that takes the
approach of teaching computers to learn for themselves, rather than
teaching computers all that they need to know. ML is the foundation for
modern AI. ML focuses on identifying and making sense of the patterns
and structures in data.

ML is about machines’ reasoning and decision-making using software that

learns from past experiences. ML allows computers to consistently perform
repetitive and well-defined tasks that are difficult for humans to
accomplish. Over the past few years, machine learning algorithms have
proved that computers can learn tasks that are tremendously complicated
for machines, demonstrating that ML can be employed in a wide range of
scenarios and industries.

AI is now being embedded into the software you use today, sometimes
without us realizing it. For example, Microsoft PowerPoint has a feature
called Design Ideas that offers suggestions for themes and layouts for
slides, and Microsoft Word offers suggestions to rewrite sentences to

93
 improve clarity.

2. Identifying features of common AI workloads

Artificial Intelligence is software that mimics human behaviours and

capabilities. Today, software can use AI to automatically detect and predict
actions that machines, and humans, should take.

Microsoft Azure provides a set of services for Artificial Intelligence and

machine learning that you can utilize to create your own intelligent
solutions. Microsoft Azure AI Fundamentals is a certification that requires
you to have entry-level knowledge of AI and ML concepts and knowledge of
the related Microsoft Azure services.

3. This skill covers how to:

a) Describe Azure services for AI and ML

b) Understand Azure Machine Learning
c) Understand Azure Cognitive Services
d) Describe the Azure Bot Service
e) Identify common AI workloads

a) Describe Azure services for AI and ML

There is a wide and rapidly growing series of services in Azure for AI and
ML

➢ Cognitive Services A set of prebuilt services that you can easily use in
your applications.
➢ Azure Bot Service A service to help create and deploy chatbots and
intelligence agents.
➢ Azure Machine Learning A broad range of tools and services that allow
you to create your own custom AI.
➢ We will be exploring some of the features and capabilities of these three
services. However, these services do not work in isolation; they utilize
many other Azure services to deliver solutions such as the following:
➢ Storage

94
 ▪ Compute
▪ Web Apps
▪ HD Insights
▪ Data Factory
▪ Cosmos DB
▪ Azure Functions
▪ Azure Kubernetes Service (AKS)

To explain how Azure services support Azure Machine Learning,

consider the scenario of a company that wants to provide recommendations
to its users. By providing personalized targeted recommendations, users will
more likely purchase more of their products and user satisfaction will
increase.

Fig 2.1 shows an example of an ML architecture to support recommendations.

Fig 2.1: example of ML architecture

b) Understand Azure Machine Learning: Azure Machine Learning is the

foundation for Azure AI. In Azure Machine Learning, you can build and
train AI models to make predictions and inferences. Training a machine
learning model requires lots of data and lots of computing resources. Azure
provides many services for preparing data and then analysing the data.

95
 cAzure ML is a platform for training, deploying, and managing machine
learning models.
c) Machine learning model types: Machine learning makes use of
algorithms to identify patterns in data and take action. The types of
machine learning models created from the outputs of the algorithms are
as follows:
a. Anomaly Detection -Finds unusual occurrences.
b. Classification- Classifies images or predicts between several
categories.
c. Clustering (including K-Means Clustering)-Discovers structures.
d. Regression- Predicts values.
d) Understand Azure Cognitive Services: Cognitive Services is a suite of
prebuilt AI services that developers can use to build AI solutions. Cognitive
Services meets common AI requirements and allow you to add AI to your
apps more quickly with less expertise. Cognitive Services are machine
learning models trained by Microsoft with massive volumes of data. While
you can build your own custom AI models to perform the same analyses,
Cognitive Services allow you to meet many AI requirements easily around
processing images and analyzing text. However, Cognitive Services only
address a subset of AI requirements. You can create your own machine
learning models to meet more complex and specific needs.
a. Cognitive Services are available as a set of REST APIs for the
following capabilities:
b. Computer vision, including images and video
c. Decision, including Anomaly Detector
d. Language, including text analysis and Language Understanding
e. Speech, including translation
f. Intelligent search, including knowledge minin

Cognitive Services have a broad and growing range of features and

capabilities. The Azure AI Fundamentals exam focuses on two of these
capabilities:

a) Image processing
b) Natural Language Processing (NLP)

96
A great example of the use of Cognitive Services is the free Seeing AI app that
uses these two capabilities. Designed for visually impaired people, this app
turns the physical environment into an audible experience, locating faces,
identifying objects, and reading documents.

Fig: The Seeing AI app

e) Describe the Azure Boot Service: The Azure Bot Service is a cloud-based
platform for developing, deploying, and managing bots. Azure Bot Services
provide the capabilities known as conversational AI. Conversational AI is
where the computer simulates a conversation with a user or customer.
Conversational AI has extended beyond simple chatbots to intelligence
agents and virtual assistants like Cortana.

There are two conversational AI services included in the Azure AI

▪ QnA Maker- A tool to build a bot using existing support and other
documentation.
▪ Azure Bot Service Tools to build, test, deploy, and manage bots.

Both QnA Maker and the Azure Bot Service leverage the Language
Understanding (LUIS) service in Cognitive Services.

97
 Identify common AI workloads

There are many use cases for AI. Here we will look at some common AI
workloads, describing their features and providing some examples for their
use.

f) Conversational AI: Conversational AI is the process of building AI agents

to take part in conversations with humans. Conversational AI is commonly
experienced by humans as chatbots on websites and other systems. AI
agents (bots) engage in conversations (dialogs) with human users. Bots use
natural language processing to make sense of human input, identify the
actions the human wants to perform, and identify the entity on which the
actions are to be performed. Bots can prompt the human for the
information required to complete a transaction.

There are three common types of bot that you may encounter:

▪ Webchat
▪ Telephone voice menus (IVR)
▪ Personal Digital Assistants

3. Describe fundamental principles of machine learning on Azure

Machine learning (ML) is the current focus of AI in computer science.

Machine learning focuses on identifying and making sense of the patterns
and structures in data and using those patterns in software for reasoning
and decision making. ML uses past experiences to make future
predictions.

ML allows computers to consistently perform repetitive and well-defined

tasks that are difficult to accomplish for humans. Over the past few years,
machine learning algorithms have proved that computers can learn tasks
that are tremendously complicated for machines and have demonstrated
that ML can be employed in a wide range of scenarios and industries.

This chapter explains machine learning algorithms such as clustering,

classification, and regression. The chapter then explains how machine

98
 learning works in terms of organizing datasets and applying algorithms to
train a machine learning model. The chapter then looks at the process of
building a machine learning model and the tools available in Azure.

4. Describe features of computer vision workloads on azure : Computer

vision is an area of artificial intelligence where software systems are
designed to perceive the world visually using cameras, images, and video.
The challenge here is that humans and computers see different things
when they look at the same object. Where a human sees an apple (object),
a machine sees an array of pixel values (image color data). To give machines
a higher-level understanding of what the image data represents, we use
pixel values as numeric features to train a machine learning model. This
model behaves like a pattern-detection function, mapping computer-
friendly features (pixel values) into human-friendly labels (objects,
attributes) in a probabilistic manner. When we feed an input image to this
model, it can now predict a relevant label with an associated confidence
value. In some sense, we have taught the computer to “see” the image the
way humans would.
5. Describe features of natural language processing (NLP) workloads on
Azure: Natural language processing (NLP) has many uses: sentiment
analysis, topic detection, language detection, key phrase extraction, and
document categorization.

Specifically, you can use NLP to:

• Classify documents. For instance, you can label documents as sensitive

or spam.
• Do subsequent processing or searches. You can use NLP output for
these purposes.
• Summarize text by identifying the entities that are present in the
document.
• Tag documents with keywords. For the keywords, NLP can use
identified entities.
• Do content-based search and retrieval. Tagging makes this
functionality possible.

99
• Summarize a document's important topics. NLP can combine identified
entities into topics.
• Categorize documents for navigation. For this purpose, NLP uses
detected topics.
• Enumerate related documents based on a selected topic. For this
purpose, NLP uses detected topics.
• Score text for sentiment. By using this functionality, you can assess the
positive or negative tone of a document.

100