Six Steps To Master Machine Learning With Data Preparation
Six Steps To Master Machine Learning With Data Preparation
Data Preparation
To prepare data for both analytics and machine learning initiatives teams can accelerate machine
learning and data science projects to deliver an immersive business consumer experience that
accelerates and automates the data-to-insight pipeline by following six critical steps.
Organizations today continue to look for ways to prepare data quickly and more accurately to
solve their data challenges and enable machine learning (ML). But before bringing your data into
a machine learning model or any other analytics project, it’s important to ensure that it is clean,
consistent, and accurate. Because much of today’s analytics is dependent on the context of the
data, the task is best done by those closest to what the data actually represents; the business
domain expert who can apply hunches, theories, and business knowledge to the data.
Unfortunately, business users don’t usually come equipped with data science skills so bridging
that gap can make the difference between gaining value from your data quickly. As result, many
are applying data preparation (DP) to help data scientists and ML practitioners rapidly prepare
and annotate their enterprise data to extend the value of the data across the enterprise for analytic
workloads.
How data collection and preparation are the foundation for trusted ML models
To create a successful machine learning model, it is imperative that an organization has the
ability to train, test, and validate them prior to deploying into production. Data preparation
technology is being used to create the clean and annotated foundation needed for today’s modern
machine learning yet, good DP historically takes more time than any other part of the machine
learning process.
Reducing the time necessary for data preparation has become increasingly important, as it leaves
more time to test, tune, and optimize models to create greater value. To prepare data for both
analytics and machine learning initiatives teams can accelerate machine learning and data
science projects to deliver an immersive business consumer experience that accelerates and
automates the data-to-insight pipeline by following six critical steps:
This is the by far the essential first step as it addresses common challenges, including:
However, when considering a DP solution, make sure it can combine multiple files into one
input, such as when you have a collection of files representing daily transactions, but your
machine learning model needs to ingest a year of data. Also, be sure to have a contingency plan
in place for overcoming problems associated with sampling and bias in your data set and your
machine learning model.
Once the data is collected, it’s time to assess the condition of it, including looking for trends,
outliers, exceptions, incorrect, inconsistent, missing, or skewed information. This is important
because your source data will inform all of your model’s findings, so it is critical to be sure it
does not contain unseen biases. For example, if you are looking at customer behavior nationally,
but only pulling in data from a limited sample, you might miss important geographic regions.
This is the time to catch any issues that could incorrectly skew your model’s findings, on the
entire data set, and not just on partial or sample data sets.
The next step in great data preparation is to ensure your data is formatted in a way that best fits
your machine learning model. If you are aggregating data from different sources, or if your data
set has been manually updated by more than one stakeholder, you’ll likely discover anomalies in
how the data is formatted (e.g. USD5.50 versus $5.50). In the same way, standardizing values in
a column, e.g. State names that could be spelled out or abbreviated) will ensure that your data
will aggregate correctly. Consistent data formatting takes away these errors so that the entire data
set uses the same input formatting protocols.
Here, start by having a strategy for dealing with erroneous data, missing values, extreme values,
and outliers in your data. Self-service data preparation tools can help if they have intelligent
facilities built in to help match data attributes from disparate datasets to combine them
intelligently. For instance, if you have columns for FIRST NAME and LAST NAME in one
dataset and another dataset has a column called CUSTOMER that seem to hold a FIRST and
LAST NAME combined, intelligent algorithms should be able to determine a way to match these
and join the datasets to get a singular view of the customer.
For continuous variables, make sure to use histograms to review the distribution of your data and
reduce the skewness. Be sure to examine records outside an accepted range of value. This
“outlier” could be an inputting error, or it could be a real and meaningful result that could inform
future events as duplicate or similar values could carry the same information and should be
eliminated. Similarly, take care before automatically deleting all records with a missing value, as
too many deletions could skew your data set to no longer reflect real-world situations.
This step involves the art and science of transforming raw data into features that better represent
a pattern to the learning algorithms. For example, data can be decomposed into multiple parts to
capture more specific relationships, such as analyzing sales performance by the day of the week,
not only the month or year. In this situation, segregating the day as a separate categorical value
from the date (e.g. “Mon; 06.19.2017”) may provide the algorithm with more relevant
information.
The final step is to split your data into two sets; one for training your algorithm, and another for
evaluation purposes. Be sure to select non-overlapping subsets of your data for the training and
evaluation sets in order to ensure proper testing. Invest in tools that provide versioning and
cataloging of your original source as well as your prepared data for input to machine learning
algorithms, and the lineage between them. This way, you can trace the outcome of your
predictions back to the input data to refine and optimize your models over time.
Data preparation has long been recognized for helping business leaders and analysts to ready and
prepare the data needed for analytics, operations, and regulatory requirements. Self-service data
preparation that runs on Amazon Web Services (AWS) and Azure takes it to the next level by
leveraging many valuable attributes of a cloud-based environment.
As a result, business users who are closest to the data and most knowledgeable about its business
context, can prepare data sets quickly and accurately, with the help of built-in intelligence and
smart algorithms. They can work within an intuitive, visual application to access, explore, shape,
collaborate and publish data with clicks, not code, with complete governance and security. IT
professionals are able to maintain the scale of data volumes and variety across both enterprise
and cloud data sources to support business scenarios for immediate and repeatable data service
needs.
Solutions like DP solve many data challenges and enable ML and data science workflows that
enhance applications with machine intelligence. More importantly, it enables them to transform
data into information on-demand to empower every person, process, and system in the
organization to be more intelligent.
Bio: David Levinger is VP Dev and Cloud Operations at Paxata, the pioneer and leader in
enterprise-grade self-service data preparation for analytics. To learn more visit www.paxata.com
or engage with the company on Twitter, LinkedIn, Facebook, or YouTube.
Introduction
Not so long ago, using the pivot tables option in Excel was the upper limit of my skills with
numbers and the word python was more likely to make me think about a dense jungle or a nature
program on TV than a tool to generate business insights and create complex solutions.
It took me ten months to leave that life behind and start feeling like I belonged to the exclusive
world of people who can tell their medians from their means, their x-bars from the neighborhood
pub, and who know how to teach machines what they need to learn.
The transformation process was not easy and demanded hard work, lots of time, dedication and
required plenty of help along the way. It also involved well over hundreds of hours of “studying”
in different forms and an equal amount of time practicing and applying all that was being learnt.
In short, it wasn’t easy to transform from being data dumb to a data nerd, but I managed
to do so while going through a terribly busy work schedule as well as being a dad to a one-
year old.
The point of this article is to help you if you are looking to make a similar transformation but do
not know where to start and how to proceed from one step to the next. If you are interested in
finding out, read on to get an idea about the topics you need to cover and also develop an
understanding of the level of expertise you need to build at each stage of the learning process.
There are plenty of great online and offline resources to help you master each of these steps, but
very often, the trouble for the uninitiated can be in figuring out where to start and where to
finish. I hope spending the next ten to fifteen minutes going through this article will help solve
that problem for you.
And finally, before proceeding any further, I would like to point out that I had a lot of help in
making this transformation. Right at the end of the article, I will reveal how I managed to
squeeze in so much learning and work in a matter of ten months. But that’s for later.
For now, I want to give you more details about the nine steps that I had to go through in my
transformation process.
Spend a couple of weeks enhancing your “general knowledge” about the field of data science
and machine learning. You may already have ideas and some sort of understanding about what
the field is, but if you want to become an expert, you need to understand the finer details to a
point where you can explain it in simple terms to just about anyone.
Suggested topics:
What is Analytics?
What is Data Science?
What is Big Data?
What is Machine Learning?
What is Artificial Intelligence?
How are the above domains different from each other and related to each other?
How are all of the above domains being applied in the real world?
Write a blog post telling readers how to answer these questions if asked in an interview
Step 2: Learn some Statistics
I have a confession to make. Even though I feel like a machine learning expert, I do not feel that
I have any level of expertise in statistics. Which should be good news for people who struggle
with concepts in statistics as much as I do, as it proves that you can be a data scientist without
being a statistician. Having said that, you cannot ignore statistical concepts – not in machine
learning and data science!
So what you need to do is to understand certain concepts and know when they may be applied or
used. If you can also completely understand the theory behind these concepts, give yourself a
few good pats on your back.
Suggested topics:
Data structures, variables and summaries
Sampling
The basic principles of probability
Distributions of random variables
Inference for numerical and categorical data
Linear, multiple and logistic regression
Create a list of references with the easiest to understand explanation that you found for each
topic and publish them in a blog. Add a list of statistics related questions that one may be
expected to answer in a data science interview
Step 3: Learn Python or R (or both) for data analysis
Programming turned out to be easier to learn, more fun and more rewarding in terms of the
things it made possible, than I had ever imagined. While mastering a programming language
could be an eternal quest, at this stage, you need to get familiar with the process of learning a
language and that is not too difficult.
Both Python and R are very popular and mastering one can make it quite easy to learn the other.
I started with R and have slowly started using Python for doing similar tasks as well.
Suggested topics:
Extract a table from a website, modify it to compute new variables, and create graphs
summarizing the data
In the first cricket test match ever played (see scorecard), Australian Charles Bannerman scored
67.35% (165 out of 245) of his team’s total score, in the very first innings of cricket’s history.
This remains a record in cricket at the time of writing, for the highest share of the total score by a
batsman in an innings of a test match.
What makes the innings even more remarkable is that the other 43 innings in that test match had
an average of only 10.8 runs an innings, with only about 40% of all batsmen registering a score
of ten or more runs. In fact, the second highest score by an Australian in the match was 20 runs.
Given that Australia won the match by 45 runs, we can say with conviction that Bannerman’s
innings was the most important contributor to Australia’s win.
Just like we were able to build this story from the scorecard of the test match, exploratory data
analysis is about studying data to understand the story that is hidden beneath it, and then sharing
the story with everyone.
Personally, I find this phase of a data project the most interesting, which is a good thing as quite
a lot of the time in a typical project could be expected to be taken up by exploratory data
analysis.
Topics to cover:
Project output:
Let’s say we had data for all the countries in the world across many parameters ranging from
population, to income, to health, to major industries and more. Now suppose we wanted to find
out which countries are similar to each other across all these parameters. How do we go about
doing this, when we have to compare each country with all the others, across over 50 different
parameters?
That is where unsupervised machine learning algorithms come in. This is not the time to bore
you with details about what these are all about, but the good news is that once you reach this
stage, you have moved on into the world of machine learning and are already in elite company.
Topics to cover:
K-means clustering
Association rules
Milestone exercise:
Practice K-means clustering on 3 different datasets from different industries or interest areas
Step 6: Create supervised learning models
If you had data about millions of loan applicants and their repayment history from the past, could
you identify an applicant who is likely to default on payments, even before the loan is approved?
Given enough prior data, could you predict which users are more likely to respond to a digital
advertising campaign? Could you identify if someone is more likely to develop a certain disease
later in their life based on their current lifestyle and habits?
Supervised learning algorithms help solve all these problems and a lot more. While there are a
plethora of algorithms to understand and master, just getting started with some of the most
popular ones will open up a world of new possibilities for you and the ways in which you can
make data useful for an organization.
Topics to cover:
Logistic regression
Classification trees
Ensemble models like Bagging and Random Forest
Supervised Vector Machines
You have not really started with creating models till you have done this:
Step 7: Understand Big Data Technologies
Many of the machine learning models in use today have been around for decades. The reason
why these algorithms are only finding applications now, is that we finally have access to
sufficiently large amounts of data, that can be supplied to these algorithms for them to be able to
come up with useful outputs.
Data engineering and architecture is a field of specialization in itself, but every machine learning
expert must know how to deal with big data systems, irrespective of their specialization within
the industry.
Understanding how large amounts of data can be stored, accessed and processed efficiently is
important to being able to create solutions that can be implemented in practice and are not just
theoretical exercises.
I had approached this step with a real lack of conviction, but as I soon found out, it was driven
more by the fear of the unknown in the form of Linux interfaces than any real complexity in
finding my way around a Hadoop system.
Topics to cover:
Upload data, run processes and extract results after installing a local version of Hadoop or Spark
on your system
Step 8: Explore Deep Learning Models
Deep learning models are helping companies like Apple and Google create solutions like Siri or
the Google Assistant. They are helping global giants test driverless cars and suggesting best
courses of treatment to doctors.
Machines are able to see, listen, read, write and speak thanks to deep learning models that are
going to transform the world in many ways, including significantly changing the skills required
for people to be useful to organizations.
Getting started with creating a model that can tell the image of a flower from a fruit may not
immediately help you start building your own driverless car, but it will certainly help you start
seeing the path to getting there.
Topics to cover:
Milestone exercise:
Create a model that can correctly identify pictures of two of your friends or family members
By now you are almost ready to unleash yourself to the world as a machine learning pro, but you
need to showcase all that you have learnt before anyone else will be willing to agree with you.
The internet presents glorious opportunities to find such projects. If you have been diligent about
the previous eight steps, chances are that you would already know how to find a project that will
excite you, be useful to someone, as well as help demonstrate your knowledge and skills.
Topics to cover:
Milestone exercise:
Get in touch with a stakeholder who will be interested in your report and share your findings wit
h them and get feedback
End Notes
Machine learning and artificial intelligence is a set of skills for the present and future. It is also a
field where learning will never cease and very often you may have to keep running to stay in the
same place, as far as being equipped with the most in-demand skills is concerned.
However, if you start the journey well, you will be able to understand how to go about taking the
next step in your learning path. As you must have gathered by now, starting the journey well is a
pretty challenging exercise in itself. If you choose to start upon it, I hope this article will have
been of some help to you and I wish you the very best.
Finally, I will confess that I got a lot of help with my ten-month transition. The reason I was able
to cover so much ground in this amount of time, along with a busy schedule at work and home,
was that I enrolled for the Post Graduate Program in Data Science and Machine Learning offered
by Jigsaw Academy and Graham School, University of Chicago.
Investing in the course helped in keeping my learning hours focused, created external pressure
that ensured that I was finding time for it irrespective of whatever else was going on in life, and
gave me access to experts in the form of faculty and a great peer group through other students.
Transforming from being non-technical to someone who is comfortable with the machine
learning world has already opened up many new doors for me. Whatever path you choose to
make this transformation, you can do so with the assurance that going through the rigor will reap
rewards for a long time and will banish any fears of becoming irrelevant in tomorrow’s
economy.
Madhukar Jha believes that great digital experiences are created by concocting a perfect mix of
data driven insights, understanding of behavioural drivers, a design thinking approach, and
cutting edge technology. He applies this philosophy to help businesses make world class
products, run campaigns that rock and tell compelling stories.
Maher
Follow
Feb 15, 2019 · 4 min read
There is a 5-Step shortcut that you can do to be able to solve machine learning problems right
away, As a beginner, you can take this path at first if you want to get something done with
machine learning.
And then you can take the 10-steps path to be a data scientist or a more advanced machine
learning engineer.
Just keep this in mind,
— Improve your mindset and your way of thinking, This is much more important than learning
how to use the tools.
— There is no fixed track to achieve your goal and become a data scientist, So you’re free to
change any step or do what you think would be right for you.
Read The Catalog
I use these tools a lot when I start solving any machine learning problem, this gives me a
baseline accuracy to try to improve upon, It helps me get more familiar with the dataset I have
and give me some insights.
1. Master python or any other language like R but I recommend you start with Python,
there’re a lot of resources for you on the web that is free, I’d suggest you go with a trial
and error approach along with some reading.
2. Learn how to use Numpy and Scipy for math, Pandas and Matplotlib and seaborn.
Numpy and Scipy, are Mathematical libraries.
Pandas is a python library for data manipulation and analysis.
Matplotlib and Seaborn are libraries that help you visualize the Data.
3. Read the article Introduction to Machine learning: Top-down approach, It’ll give you
a smooth introduction to the machine learning world.
4. Read about Scikit-learn, this step is the actual catalog reading, scikit-learn is the toolset
you’ll use to solve the problems, you don't have to learn everything in the library just
learn to implement one or two models and read about the others.
Scikit-learn is a python library that has a lot of already implemented models that are
black boxes that you can use to train and make predictions with directly, and you can
even tune the model’s parameters to suit your problem and get more accuracy.
5. Read Chapter 2 in the book Hands-On Machine Learning with Scikit-Learn &
TensorFlow
If you’re not interested in data science yet then you should not continue aiming for data science,
But if you ARE interested, I recommend that you take the second path.
Now, Let’s be Professional
The most important thing that will shape how good you are as a data scientist is that you should
always be up to date with the new discoveries out in the field.
I recommend that you read a lot of papers, follow a lot of publications and writers, and engage
with them, reach out to me for any questions! We can benefit each other and build a caring
community.
Foundations
Beginner
Understand ML Algorithms
ML + Weka (no code)
ML + Python (scikit-learn)
ML + R (caret)
Time Series Forecasting
Intermediate
Code ML Algorithms
XGBoost Algorithm
Imbalanced Classification
Deep Learning (Keras)
Better Deep Learning
Advanced
My best advice for getting started in machine learning is broken down into a 5-step process:
Step 1: Adjust Mindset. Believe you can practice and apply machine learning.
o What is Holding you Back From Your Machine Learning Goals?
o Why Machine Learning Does Not Have to Be So Hard
o How to Think About Machine Learning
o Find Your Machine Learning Tribe
Step 2: Pick a Process. Use a systemic process to work through problems.
o Applied Machine Learning Process
Step 3: Pick a Tool. Select a tool for your level and map it onto your process.
o Beginners: Weka Workbench.
o Intermediate: Python Ecosystem.
o Advanced: R Platform.
o Best Programming Language for Machine Learning
Step 4: Practice on Datasets. Select datasets to work on and practice the process.
o Practice Machine Learning with Small In-Memory Datasets
o Tour of Real-World Machine Learning Problems
o Work on Machine Learning Problems That Matter To You
Step 5: Build a Portfolio. Gather results and demonstrate your skills.
o Build a Machine Learning Portfolio
o Get Paid To Apply Machine Learning
o Machine Learning For Money
Many of my students have used this approach to go on and do well in Kaggle competitions and
get jobs as Machine Learning Engineers and Data Scientists.
To have skill at applied machine learning means knowing how to consistently and reliably
deliver high-quality predictions on problem after problem. You need to follow a systematic
process.
Below is a 5-step process that you can follow to consistently achieve above average results on
predictive modeling problems:
Step 1: Define your problem.
o How to Define Your Machine Learning Problem
Step 2: Prepare your data.
o How to Prepare Data For Machine Learning
o How to Identify Outliers in your Data
o Improve Model Accuracy with Data Pre-Processing
o Discover Feature Engineering
o An Introduction to Feature Selection
o Tactics to Combat Imbalanced Classes in Your Machine Learning Dataset
o Data Leakage in Machine Learning
Step 3: Spot-check algorithms.
o How to Evaluate Machine Learning Algorithms
o Why you should be Spot-Checking Algorithms on your Machine Learning
Problems
o How To Choose The Right Test Options When Evaluating Machine Learning
Algorithms
o A Data-Driven Approach to Choosing Machine Learning Algorithms
Step 4: Improve results.
o How to Improve Machine Learning Results
o Machine Learning Performance Improvement Cheat Sheet
o How To Improve Deep Learning Performance
Step 5: Present results.
o How to Use Machine Learning Results
o How to Train a Final Machine Learning Model
o How To Deploy Your Predictive Model To Production
Below is the 3 step process that you can use to get up-to-speed with probability for machine
learning, fast.
You can see all of the tutorials on probability here. Below is a selection of some of the most
popular tutorials.
Probability Foundations
Bayes Theorem
Probability Distributions
Information Theory
Below is the 3 step process that you can use to get up-to-speed with statistical methods for
machine learning, fast.
You can see all of the statistical methods posts here. Below is a selection of some of the most
popular tutorials.
Summary Statistics
Resampling Methods
Estimation Statistics
Below is the 3 step process that you can use to get up-to-speed with linear algebra for machine
learning, fast.
Step 1: Discover what Linear Algebra is.
o Basics of Mathematical Notation for Machine Learning
o A Gentle Introduction to Linear Algebra
Step 2: Discover why Linear Algebra is important for machine learning.
o 5 Reasons to Learn Linear Algebra for Machine Learning
o 10 Examples of Linear Algebra in Machine Learning
o Linear Algebra for Machine Learning
Step 3: Dive into Linear Algebra topics.
o Linear Algebra for Machine Learning Mini-Course
o Linear Algebra for Machine Learning (my book)
You can see all linear algebra posts here. Below is a selection of some of the most popular
tutorials.
Matrices
Vectors
Introduction to Vectors
Introduction to Vector Norms
Matrix Factorization
You need to know what algorithms are available for a given problem, how they work, and how
to get the most out of them.
Here’s how to get started with machine learning algorithms:
You can see all machine learning algorithm posts here. Below is a selection of some of the most
popular tutorials.
Linear Algorithms
Gradient Descent
Linear Regression
Logistic Regression
Linear Discriminant Analysis
Nonlinear Algorithms
Ensemble Algorithms
It has a graphical user interface meaning that no programming is required and it offers a suite of
state of the art algorithms.
You can see all Weka machine learning posts here. Below is a selection of some of the most
popular tutorials.
Below are the steps that you can use to get started with Python machine learning:
You can see all Python machine learning posts here. Below is a selection of some of the most
popular tutorials.
It’s popular because of the large number of techniques available, and because of excellent
interfaces to these methods such as the powerful caret package.
You can see all R machine learning posts here. Below is a selection of some of the most popular
tutorials.
Data Preparation in R
Learning via coding is the preferred learning style for many developers and engineers.
Here’s how to get started with machine learning by coding everything from scratch.
You can see all of the Code Algorithms from Scratch posts here. Below is a selection of some of
the most popular tutorials.
Prepare Data
Linear Algorithms
Algorithm Evaluation
Nonlinear Algorithms
Many datasets contain a time component, but the topic of time series is rarely covered in much
depth from a machine learning perspective.
Data Preparation Tutorials
Forecasting Tutorials
How to Make Baseline Predictions for Time Series Forecasting with Python
How to Check if Time Series Data is Stationary with Python
How to Create an ARIMA Model for Time Series Forecasting with Python
How to Grid Search ARIMA Model Hyperparameters with Python
How to Work Through a Time Series Forecast Project
It is popular because it is being used by some of the best data scientists in the world to win
machine learning competitions.
You can see all XGBoosts posts here. Below is a selection of some of the most popular tutorials.
XGBoost Basics
XGBoost Tuning
Imbalanced Classification
Imbalanced classification refers to classification tasks where there are many more examples for
one class than another class.
These types of problems often require the use of specialized performance metrics and learning
algorithms as the standard metrics and methods are unreliable or fail completely.
You can see all Imbalanced Classification posts here. Below is a selection of some of the most
popular tutorials.
Performance Measures
Cost-Sensitive Algorithms
Advanced Methods
State-of-the-art results are coming from the field of deep learning and it is a sub-field of machine
learning that cannot be ignored.
You can see all deep learning posts here. Below is a selection of some of the most popular
tutorials.
Background
Multilayer Perceptrons
5 Step Life-Cycle for Neural Network Models in Keras
How to Grid Search Hyperparameters for Deep Learning Models in Python With Keras
Save and Load Your Keras Deep Learning Models
Display Deep Learning Model Training History in Keras
Dropout Regularization in Deep Learning Models With Keras
Time Series Prediction with LSTM Recurrent Neural Networks in Python with Keras
Understanding Stateful LSTM Recurrent Neural Networks in Python with Keras
Text Generation With LSTM Recurrent Neural Networks in Python with Keras
There are standard techniques that you can use to improve the learning, reduce overfitting, and
make better predictions with your deep learning model.
Here’s how to get started with getting better deep learning performance:
You can see all better deep learning posts here. Below is a selection of some of the most popular
tutorials.
Better Learning (fix training)
You can see all LSTM posts here. Below is a selection of some of the most popular tutorials
using LSTMs in Python with the Keras deep learning library.
LSTM Behaviour
Here’s how to get started with deep learning for natural language processing:
You can see all deep learning for NLP posts here. Below is a selection of some of the most
popular tutorials.
Bag-of-Words Model
Language Modeling
Text Summarization
Text Classification
Word Embeddings
What are Word Embeddings?
How to Develop Word Embeddings in Python with Gensim
How to Use Word Embedding Layers for Deep Learning with Keras
Photo Captioning
Text Translation
Computer vision is not solved, but to get state-of-the-art results on challenging computer vision
tasks like object detection and face recognition, you need deep learning methods.
Here’s how to get started with deep learning for computer vision:
Step 1: Discover what deep learning for Computer Vision is all about.
o What is Computer Vision?
o What is the Promise of Deep Learning for Computer Vision?
Step 2: Discover standard tasks and datasets for Computer Vision.
o 9 Applications of Deep Learning for Computer Vision
o How to Load and Visualize Standard Computer Vision Datasets With Keras
o How to Develop and Demonstrate Competence With Deep Learning for
Computer Vision
Step 3: Discover how to work through problems and deliver results.
o How to Get Started With Deep Learning for Computer Vision (7-Day Mini-
Course)
o Deep Learning for Computer Vision (my book)
You can see all deep learning for Computer Vision posts here. Below is a selection of some of
the most popular tutorials.
Image Classification
Object Recognition
Methods such as MLPs, CNNs, and LSTMs offer a lot of promise for time series forecasting.
Here’s how to get started with deep learning for time series forecasting:
Step 1: Discover the promise (and limitations) of deep learning for time series.
o The Promise of Recurrent Neural Networks for Time Series Forecasting
o On the Suitability of Long Short-Term Memory Networks for Time Series
Forecasting
o Results From Comparing Classical and Machine Learning Methods for Time
Series Forecasting
Step 2: Discover how to develop robust baseline and defensible forecasting models.
o Taxonomy of Time Series Forecasting Problems
o How to Develop a Skillful Machine Learning Time Series Forecasting Model
Step 3: Discover how to build deep learning models for time series forecasting.
o How to Get Started with Deep Learning for Time Series Forecasting (7-Day Mini-
Course)
o Deep Learning for Time Series Forecasting (my book)
You can see all deep learning for time series forecasting posts here. Below is a selection of some
of the most popular tutorials.
Models Types
GANs are an exciting and rapidly changing field, delivering on the promise of generative models
in their ability to generate realistic examples across a range of problem domains, most notably in
image-to-image translation tasks.
Here’s how to get started with deep learning for Generative Adversarial Networks:
You can see all Generative Adversarial Network tutorials listed here. Below is a selection of
some of the most popular tutorials.
GAN Fundamentals
If you still have questions and need help, you have some options:
Ebooks: I sell a catalog of Ebooks that show you how to get results with machine
learning, fast.
o Machine Learning Mastery EBook Catalog
Blog: I write a lot about applied machine learning on the blog, try the search feature.
o Machine Learning Mastery Blog
Frequently Asked Questions: The most common questions I get and their answers
o Machine Learning Mastery FAQ
Contact: You can contact me with your question, but one question at a time please.
o Machine Learning Mastery Contact