BUSINESS ANALYTICS

LESSON 1
Business - practice of making one's living
by engaging in commerce. [Oxford]
Analytics - method of logical analysis EVOLUTION OF BUSINESS ANALYTICS
[Merriam webster]
1800: F. Taylor created first system of
BUSINESS ANALYTICS business analytics: Scientific Management
 process of transforming data into of USA.
insights to improve business 1950: IBM invented hard disk drive to
decisions. store data; business relied on manual
 Data management, data record keeping & simple statistical method
visualization, predictive modeling, 1950-1980: BI solutions was based on
data mining, forecasting simulation, databases & data warehousing; MIS & DSS
optimization → tools to create introduced
insights from data. 1980-2000: BI rose - integrated data from
various sources; Data warehousing
emerged at large scale to store & analyze
data
2000-2010: organizations recognized the
need for BI; data generation rose in the
internet, social media & IoT devices;
2010-Present: AI & machine learning; real
time analytics developed; with advanced
AI-powered insight
BUSINESS ANALYTICS & BUSINESS Future: will continue to rise & improve
INTELLIGENCE
BA is a subdivision of BI IMPORTANCE OF BUSINESS ANALYTICS
Business Business  transform raw data into more valuable
Intelligence Analytics inputs to leverage information in
Definition Used to analyze Used to decision making.
historical and analyze
present data to historical
 With Business Analytics tools, a more
understand and data to profound understanding of primary
drive current drive and secondary data emerging from
business current and their activities.
operations future  helps businesses refine their
business
procedures further and be more
Usage Present business Future
operations business productive.
operations  To stay competitive, companies need
Application Suitable for large Suitable for to be ahead of their competitors
companies all  assist their decision making in
companies improving efficiency as well as
Tools PowerBI, SAP, Microsoft
generating more profits.
QlikSense, etc. Office
(Excel),
Google IMPORTANCE OF BUSINESS ANALYTICS
Analytics,  Informed Decision-making
Looker, etc.  Improved efficiency & operations
 BUSINESS ANALYTICS
 Better customer insights & a. recommends specific
personalization solutions for businesses to
 Competitive advantage drive their growth
 Financial management & profitability forward.
 Risk management b. What should we do?
 Improved marketing strategy 5. Cognitive
 Enhanced forecasting & planning a. searches through the data
 Real-Time Monitoring and Decision- that exists in its
Making knowledge base to find
 Continuous Improvement and solutions that make sense
Innovation for the questions posed.
 Employee Performance and Workforce b. How can machines assist
Optimization in decision-making?
6. Real-time
BENEFITS OF BUSINESS ANALYTICS a. use of data and related
 More Informed Decision-Making resources for analysis as
 Better: better view of information soon as it enters the
 Faster: prompts organization to system.
take action faster b. What is happening right
 Greater Revenue now?
 May experience higher financial
returns WHO USES BUSINESS ANALYTICS?
 Improved Operational Efficiency  Top Management
 Fine tunes business processes &  CEOs, CFOs, Department Heads,
operations BODs
 Enhanced customer service  Data & Analytics Professionals
 Data/Business Analysts, BI
TYPES OF BUSINESS ANALYTICS Analysts, Data Scientists
TECHNIQUES  Functional Managers
1. Descriptive  Managers; analysts (in specific
a. describes the past or field)
present situation of the  IT & Cybersecurity Group
organization's activities.  IT Managers; Cybersecurity
b. What happened? Experts & analysts
2. Diagnostic  Industry Specific Users
a. discovers factors or  Managers & Analysts in other
reasons for past or current areas
performance.
b. Why did it happen? LESSON 2
3. Predictive Importance of Data in Business
a. predicts figures and The Data in the Organization
results using a  Data - foundational asset of
combination of business organization for decision making.
analytics tools.
b. What will happen? What happens without data?
4. Prescriptive
 BUSINESS ANALYTICS
o “Without data you only have an  Uncover hidden insights;
opinion; with data you can have a anticipate trends, optimize
well-founded opinion” - W. Edwards operations, enhance decision
Deming making
o With technology and people, data is 5. Data Visualization
interrelated  Makes insights digestible;
 Thru complex processes that empowers team to act quickly on
enable efficient and agile critical business trends
strategic decision making, in 6. Reporting and Action
response to disruptive  Continuous reporting & real time
changes insights → make business stay
ahead, maximize performance
The Data in the Organization
 The value of data in the business Strategic Value of Data
world has increased significantly. 1. Informed Decision-Making
 In Decision making, it all comes down 2. Competitive Advantage
to data; therefore, data should be 3. Operational Efficiency
o Correct 4. Customer Experience &
o Reliable Personalization
o Timely (available when needed) 5. Risk Management & Compliance
o Relevant 6. Innovation & New Business
o Sufficient Models
7. Enhanced Collaboration &
Strategic Value of Data Knowledge Sharing

Competitive Advantage of Data

Utilization
 understand their customers,
 make better decisions
 cut costs, and
 improve their products and
BI PROCESS services
1. Data Collection
 Foundation of having a clear view  “A well utilized data is the secret
of organization → make informed weapon of the organization in
decisions order to stay ahead in the
2. Data ETL business”
 Ensures data integrity; prepares
business to generate reliable Role of Data in Decision Making
insights Data-driven decision-making (DDDM)
3. Data Storage  uses facts, metrics, and data to guide
 Not to worry about data overload strategic business decisions that align
and compliance risks with your goals, objectives, and
4. Data Analysis initiatives.
1. Enhancing Accuracy and
Objectivity
 BUSINESS ANALYTICS
2. Predictive and Prescriptive  Leadership intervention and
Analytics commitment
3. Real-Time Decision Making  Data Empowerment
4. Improving Efficiency and o Accessibility and Transparency
Productivity  Collaboration
5. Risk Management and Compliance o recognize & reward data driven
6. Enhancing Customer Experience decision
7. Supporting Strategic Planning  Value Realization
8. Facilitating Innovation o Key Metrics & KPIs
 Technology & infrastructure
Data-driven Culture o DBMS, BI, AI, hardware and other
 About humans making decisions based software
on researched data
 About using automation to collect LESSON 3
data and make real time decisions I. The Business Analytics Process
 Being a data driven organization
o Not only improves decision PROCESS:
making BUT 1. Discover/define business
o Also improves efficiency and requirements/needs
user-friendliness o what to improve?
o what problem to solve?
Advantages of data-driven organization o What data is needed/available
 Increased productivity and team o How data can be used?
effectiveness 2. Collect and explore the data
 Improved transparency and workforce o Clean data
engagement o make computations for
 Automation improves coordination missing data, remove outliers
and consistency o transform combinations of
 Data-driven cultures save money and variables
increase profits o perform descriptive statistics
to provide basic
Creating a data-driven culture
understanding of data
 Software makes a good impact but a
3. Analyze the data
partial solution
o use statistical methods
 The workplace culture must embrace
(correlation analysis,
the use of data in day-to-day
Hypothesis testing)
operations and use it in decision
o analyst find factors related to
making process
target variables
 Creating a data-driven culture requires
o analysts perform simple
a commitment.
regression analysis to see
o Trust in the data
simple predictions are made
o Trust in others
o groups are tested, too
o Commitment to work
4. Make Predictions
o analysts will model data
Elements of a data-driven culture
5. Select Optimal Solution
 BUSINESS ANALYTICS
6. Make decision & measure the o Plot data using scatter plots
outcome (identify possible correlation or
7. Update the system with the non-linearity)
results of the decision o Check all possible slices of data and
o Summarize data using appropriate
STAGE 1: DEFINING THE BUSINESS visualization and descriptive
PROBLEM statistics (mean, median, mode, SD,
 Understand range)
o what the business wants to  Will help provide a basic
improve understanding of the data
o What problem/s to solve o Analyst is already looking for
 Goals should be broken down into general patterns and actionable
smaller goals insights that can be derived to
o Relevant data to solve the goals are achieve the goal
decided by stakeholders, users with
the domain knowledge and DATA EXPLORATION
business analyst/s What is Data Exploration
 What data is available?  The first step in data analysis
 How data can be used? o Where data analysis begins
 Are there sufficient data? o Preliminary investigation of
dataset
Data Collection and Cleaning  Allowing to better
 Exploring the data understand a dataset the
 Cleaning the data team is working with
o Making computations for missing  Dataset’s characteristics,
data patterns and potential
o Remove outliers insights that may be
o Transform combinations of gleaned from it.
variables to form new variables  Teams can build
o Plot > Time series graphs > see hypotheses; make
outliers > remove outliers from informed decisions
dataset about the appropriate
 Outlier - an observation that methods to use for
deviates significantly from deeper analysis
other observations, raising  Patterns, correlations, and points for
suspicions that it was further inquiry emerge.
generated by a different  Data exploration is a crucial step
mechanism o Lays the foundation for more
 Outliers often affect advanced analytics, modelling
accuracy of model and interpretation
 When data is clean  Provides a wide-angle view of the
o Analysts will try to make better data
sense of data  Lays the groundwork for in-depth
analysis
 Data exploration vs data discovery
 BUSINESS ANALYTICS
o Data discovery follows data  Simple summaries and visualizations
exploration, taking place after generated can open line for further
the data has been prepared and inquiry without the use of normal
organized; modeling techniques
 Involves using the prepared Better data governance practices
data to answer business  Data handled must be aligned with
questions regulatory and organizational
 Include data exploration as requirements
a subset, BUT it goes  Example:
beyond the initial o Snowflake horizon
exploration phase  Has a built-in governance
o Data exploration focuses on solutions with a unified set
understanding dataset’s of compliance, security,
characteristics and patterns privacy and interoperability
and access capabilities.
BENEFITS OF DATA EXPLORATION  protects and audits
 Develop a deeper understanding of customers’ data with
data compliance, business
 Uncover insights continuity, data quality
 Better data governance practices monitoring
 Detect anomalies and other data o Microsoft Power BI
issues  Prevent data oversharing
 Check assumptions and develop  implements multiple
hypotheses functions to ensure data
governance and security,
such as encryption, access
control, sensitivity labels
Detect anomalies and other data issues
Develop a deeper understanding of data  Identifying and addressing outliers,
 The cornerstone of a successful data errors and anomalies during the data
analytics program exploration can enhance data quality
 When using data visualization, manual  By uncovering potential
analysis and other methods problems/issues with data in the early
(combined) → Team can discover new stages, may produce more accurate
sources of data, make sense of their results (thru in-depth analysis)
metadata and gauge relevance and Check assumptions and develop
quality of data at hand hypotheses
 In machine learning, data exploration  Initial assumptions aren’t always
provided essential information needed supported by the data.
for model building and feature  Investigative process (of DE) makes it
engineering possible to check assumptions
Uncover insights o Providing an opportunity to align
 Find patterns, trends and relationships what was thought to be true with
with a dataset what the data actually supports
 BUSINESS ANALYTICS
o The initial data snapshot is useful Data Exploration involves
for identifying interesting areas  Descriptive Statistics: Summarizing a
for more formal hypothesis dataset's distribution's central
testing and follow up tendency, dispersion, and shape.
 Visualization: Using charts, plots, and
Data Exploration in Business Analytics graphs to identify trends, patterns,
 Businesses use data exploration and outliers visually.
o Uncover new markets  Profiling: Understanding the structure,
o Find ways to improve existing relationships, and quality of the data.
products or services  Segmentation: Dividing the data into
o Better understand the needs meaningful subgroups based on
and wants of customers specific characteristics.
 Quick examination of data
o Leads to identifying patterns like II. MODEL BUILDING
customer purchasing behavior,
trends, border shifts Model Building in Business Analytics
o Information gathered can  involves creating predictive or
inform the more in-depth descriptive models that help
inquiry that follows businesses make data-driven
 Allowing business to fine decisions.
tune marketing strategy,  cornerstone for extracting insights,
modify products/services, making predictions, and solving
launch new complex problems.
products/services  models can forecast trends, identify
patterns, optimize operations, and
Data Exploration Tools support strategic planning
 Manual data exploration method  The process of creating effective
o Either writing scripts to analyze models is not limited to algorithms
raw data or manually filtering and data; it’s about using the right
data into spreadsheets methods.
 Automated data exploration tools
o Data visualization software Model Building Process in BA
o BI Tools 1] Problem Definition
o data preparation software  Identify business problem &
objective (forecasting,
vendors
segmentation)
o data exploration platforms
 Define the goal of the model
(prediction, classification,
optimization)
2] Data Collection & Preparation
 Collect data (internal or external)
 Clean & pre-process data
(handling missing values, outliers,
normalization)
 BUSINESS ANALYTICS
 Feature engineering (creating new o helps us evaluate the
features, encoding categorical performance of a trained model
variables) o checking whether your model
3] Model Selection meets the assumptions and
Choose appropriate type of model based requirements of the chosen
on business problem & data algorithm, and whether it fits the
characteristics data well.
 Predictive models: regression,  involves evaluating the performance
time series, forecasting and reliability of predictive models to
 Descriptive models: clustering, ensure they produce accurate and
association rules generalizable results.
 Prescriptive model: optimization,  a critical step in the modeling process
simulation to prevent overfitting/underfitting,
4] Model Training assess model performance, and build
 Split data into training and testing trust in the model's predictions.
sets. o Overfitting: the model is too
 Train the model using historical complex and captures the noise
data. and outliers in the training data,
 Use techniques like cross- but fails to generalize to new
validation to avoid overfitting data.
5] Model Evaluation o Underfitting: the model is too
 Evaluate the model using metrics simple and misses the patterns
o Classification models: and relationships in the data,
accuracy, precision, recall, F1 resulting in low performance.
score  A model should perform well in new
o Regression models: RMSE, data
MAE, R².  Various techniques are used to
o Clustering Models: validate the mode
Silhouette Score, Davies-
Bouldin Index. Model Validation Techniques
 Compare model performance and  Holdout method:
select the best model. o An in-sample validation: use of
6] Model Deployment data from the same dataset that
 Integrate the model into business was employed to develop the
processes or applications. model.
 Automate model predictions (e.g., o The dataset is then divided into
through APIs, dashboards). training set which is used to train
7] Monitoring & Maintenance the model and a hold out set which
 Continuously monitor model is used to test the performance of
performance. the model.
 Retrain the model periodically  Training, Validation, and Test
with new data if needed. Sets: Split the data into three
sets. The model is trained on
Model Validation the training set, fine-tuned on
 A process
 BUSINESS ANALYTICS
the validation set, and o is useful especially where data in
evaluated on the test set. one class is very low compared to
 Common Splits: 70-15-15 or others.
80-20 for training and testing.
o a straightforward method, but it is Model Evaluation
prone to overfitting if the holdout  process of measuring how well your
sample is small. model performs on new and unseen
 K-Fold Cross-validation: data, usually called the test or holdout
o Out sample validation: relies on set.
entirely different data from the  helps you to estimate the predictive
data used for training the mode power and generalizability of your
o The data is divided into k number model, and compare it with other
of folds. models or benchmarks.
o The model is trained on k-1 folds  To evaluate your model: choose
and tested on the fold that is left. appropriate metrics and criteria that
This is repeated k times, each time reflect the goals and objectives of your
using a different fold for testing. project, and the characteristics of your
 The data is divided into 'k' data.
subsets.
 The model is trained 'k' times, Model Validation Techniques
each time using a different Performance Metrics
subset as the test set and the  Classification Models: Accuracy,
remaining as the training set. precision, recall, F1-score, ROC-AUC.
o offers a more extensive analysis  Regression Models: Mean Absolute
than the holdout method. Error (MAE), Mean Squared Error
 Leave-One-Out Cross-validation (MSE), R² score.
(LOOCV): a form of k-fold cross  Other Metrics: Confusion matrix, lift
validation where k is equal to the and gain charts.
number of instances. Statistical Techniques
 Special case of k-fold where k  Hypothesis Testing: To check if model
is equal to the number of data improvements are statistically
points. significant.
 Only one piece of data is not  Bootstrap Methods: Generating
used to train the model. multiple samples to assess the stability
 This is repeated for each data of the model.
point.
o Unfortunately, LOOCV is also time Communicating Results
consuming when dealing with large  a crucial step in the business analytics
datasets. process
 Stratified K-Fold Cross-validation:  it transforms data-driven insights into
o k-fold cross-validation: actionable strategies that stakeholders
o each fold has the same ratio of can understand and act upon.
classes/categories as the overall o Effective communication bridges
dataset. the gap between complex
 BUSINESS ANALYTICS
analytics and business decision-
making.
1. Who is your audience?
 Top management/executives: high
level insights , strategic results and
implications
 Operational members: show practical
applications and specific action steps
to do
 Technical members: show detailed
methodologies, metrics used, and
technical findings
2. Choose the communication format
 Dashboards: real-time data
visualization
 Presentations: summarize insights
with visuals
 Reports: detailed analysis with
narratives, data and recommendations
 Data stories: narratives that guides
stakeholders through the data journey

 Use data visualization effectively

 Highlight key metrics and KPIs
o Revenue growth, costs savings, etc.
o forecast/trends, risks, opportunities
 Tell a compelling data story
o Explain the business problem and
objectives
o Show findings, trends, potential
business outcomes
o Recommend clear and data-driven
actions
 Make results actionable
o Provide recommendations, outline
and show next steps, present risk
assessment (limitations and
challenges)