Zero to hero in 4 weeks with

Machine Learning
A guide to achieving Machine Learning expertise on Azure

The cloud is the destination for an increasing number of machine learning

projects. It’s an easy, cost-effective way to experiment and scale at any level of
expertise or demand.

This learning journey will teach you how to create innovative solutions for
complex problems with Machine Learning on Azure in 4 short weeks. With just
an hour each day—think coffee-fueled morning ritual or mid-afternoon
break—you’ll be able to collaborate and build models faster with the latest
machine learning tools and frameworks. Regardless of where you are in your
journey, we can propel you to the next level and prepare you for the Azure
Data Scientist Associate certification. Let’s get started!
There’s so much to learn about Machine Learning on Azure. Don’t worry—we’ve curated an
Expertise Attain Azure Data Scientist
easy-to-understand journey to prepare you from the very basics all the way to scaling your
achieved Associate certification
ML projects. Click on the icon to jump to that week’s training.

Before you begin, click here to prepare for your journey.

Advanced
skill reached

Basics
mastered

WEEK 1 WEEK 2 WEEK 3 WEEK 4

Understand the fundamentals of Learn the essentials of building Acquire skills to operationalize Scale your skills with ML
Machine Learning ML solutions using Python ML models on Spark
Learn the core principles of ML and the basics Utilize Azure Machine Learning to create ML Practice deploying, managing, and Harness the power of Azure Databricks and
of creating ML models using tools such as models and pipelines with the Python SDK monitoring machine learning models with Apache Spark to run data science workloads
AutoML and the drag-and-drop designer. and notebooks. Azure Machine Learning. in an optimized, agile cloud environment.
 Back to Learning Journey

Make sure to complete the following tasks before you begin.

Sign up for the 30 Days to Learn It challenge and after completing all required
learning modules, you’ll be eligible to receive 50 percent off a Microsoft Certification
exam*.

Sign up for an Azure Subscription. You’ll need it for hands-on learning.

Create an Azure Machine Learning workspace with a compute instance and the ml-
basics repository cloned.

Create a trial workspace for Azure Databricks with a single node cluster.

Use the navigation bar to head back to your Learning Journey to begin your training or
click ahead to move to week 1.

*Visit Terms & Conditions to learn more.

 Back to Learning Journey
WEEK 1

Week 1
Understand the fundamentals of Machine
Learning
Machine learning models are the heart of many modern applications and services.
With a commitment to open source, Azure Machine Learning helps data scientists and
ML engineers rapidly build and deploy machine learning regardless of skill level.

This week you’ll learn the core principles of machine learning by exploring and
analyzing data with Python. To help you build on your terms, we have two areas of
focus this week: (1) how to train, evaluate, and deploy machine learning models using
common tools and frameworks, and (2) how to publish a variety of learning models
with without writing a single line of code.

Click below to access the training or click ahead to view the training modules.

Create machine learning models Create no-code predictive models

Click to begin training > with Azure Machine Learning
Click to begin training >
 Back to Learning Journey
WEEK 1

Click the box to launch each module. Once completed, check the box to easily track your progress.

Create machine learning models Create no-code predictive models with Optional resources
Azure Machine Learning To learn even more click on these
optional trainings.

30 mins 35 mins 45 mins 55 mins 1 hr 30 mins

1. Explore and analyze data 2. Train and evaluate 1. Use automated machine 2. Create a Regression Model Developer’s intro to data
with Python regression models learning in Azure ML with Azure ML designer science video series
Learn common data exploration and Know when to use regression models Identify types of ML models and use Use Azure ML designer to train a Learn the basics of data science on
analysis tasks, as well as how to use and how to train and evaluate them automated ML to train and deploy a regression model, then use it for Azure with these bite-sized videos.
Python packages. using the Scikit-Learn framework. predictive model. inferencing and deploy it as a service.

35 mins 35 mins 1 hr 50 mins 12 mins

3. Train and evaluate 4. Train and evaluate 3. Create a classification model 4. Create a Clustering Model with What is automated machine
classification models clustering models with Azure ML designer Azure ML designer learning (AutoML)?
Know when to use classification Know when to use clustering models Use Azure ML designer to train a Use Azure ML designer to train a Understand what AutoML is, how to
models and how to train and and how to train and evaluate them classification model, then use it for clustering model, then use it for works, and when to use it.
evaluate them using the Scikit-Learn using the Scikit-Learn framework. inferencing and deploy it as a service. inferencing and deploy it as a service.
framework.

1 hr 10 mins 12 mins

5. Train and evaluate deep What is Azure Machine

learning models Learning designer?
Learn the basic principles of deep Discover more about the Azure ML
learning and how to train neural Check out the optional resources to the right. Then use the navigation bar to return to designer – a visual drag-and-drop
networks using PyTorch or canvas to create ML models.
your learning journey or click ahead for week 2.
Tensorflow.
 Back to Learning Journey
WEEK 2

Week 2
Learn the essentials of building ML solutions
using Python
Trying to meet the rising demands for enterprise-ready ML solutions without the right
tools has left many in the data science community bewildered. Azure Machine Learning
fills the gap, delivering the capabilities enterprise data science teams need to be
productive from day 1 including a collaborative workspace and integrated compute
environments with support for all open source tools, frameworks, and libraries.

This week dive into the Azure Machine Learning Python SDK. We’ll start with the
essentials including how to use it to provision a workspace, train a machine learning
model, and create machine learning pipelines. You won’t want to skip this week!

Click below to access the training or click ahead to view the training modules.

Click to begin training >

 Back to Learning Journey
WEEK 2

Build AI solutions with Azure Machine Learning Optional resources

Click the box to launch each module. Once completed, check the box to easily track your progress. To learn even more click on these
optional trainings.

40 mins 45 mins 40 mins 45 mins 24 mins

1. Introduction to the Azure 2. Train a machine learning 3. Work with Data in Azure 4. Work with Compute in The Team Data Science
ML SDK model with Azure ML ML Azure ML Process (TDSP) lifecycle
See how to train, deploy, and Use Azure Machine Learning to train Create and use datastores and Find out what lifecycle the TDSP
Find out how to use cloud compute recommends you use to structure
manage machine learning models a model and register it in a datasets in Azure ML to build in Azure ML to run training your data-science projects.
with the cloud-based platform from workspace. scalable, cloud-based model training experiments at scale.
Azure ML. solutions.

55 mins 40 mins 11 mins

5. Orchestrate machine 6. Deploy real-time ML How Azure Machine Learning

learning with pipelines services with Azure ML works
Learn about the architecture of
Create, publish, and run pipelines to Learn how to register and deploy ML
Azure ML and get an understanding
train models in Azure ML. models with the Azure ML services. of how the components work
together.

2 mins read + poster download

Machine Learning Algorithm
Cheat Sheet
Choose the right algorithm from the
Check out the optional resources to the right. Then use the navigation bar to return to your designer for a predictive analytics
learning journey or click ahead for week 3. model.
 Back to Learning Journey
WEEK 3

Week 3
Acquire skills to operationalize ML models
Azure Machine Learning helps you operationalize machine learning models
irrespective of criteria stringency and encourages the use of state-of-the-
art responsible ML capabilities to understand, protect, and control your models.

This week focuses on operationalizing your machine learning models with optimal
accuracy and fairness. You’ll practice advanced skills such as publishing a batch
inference pipeline and tuning hyperparameters. Plus, you’ll learn how to detect and
mitigate unfairness and data drift to ensure continual success. What are you
waiting for - dive in!

Click below to access the training or click ahead to view the training modules.

Click to begin training >

 Back to Learning Journey
WEEK 3

Build AI solutions with Azure Machine Learning Optional resources

Click the box to launch each module. Once completed, check the box to easily track your progress. To learn even more click on these
optional trainings.

45 mins 45 mins 25 mins 45 mins 30 mins

7. Deploy batch inference 8. Tune hyperparameters with 9. Automate ML model 10. Explain machine learning GigaOm report – Delivering
pipelines with Azure ML Azure ML selection with Azure ML models with Azure ML on the vision on MLOps
Learn how to create, publish, and use See how to use Azure ML Find out how to use automated ML Understand how to explain models Download the GigaOm MLOps
batch inference pipelines with Azure hyperparameter tuning experiments in Azure ML to find the best model by calculating and interpreting report which includes best practices
Machine Learning. to optimize model performance. for your data. feature importance. for effective ML implementation.

45 mins 40 mins 40 mins 4 mins

11. Detect and mitigate 12. Monitor models with 13. Monitor data drift with Whitepaper - Drive efficiency
unfairness in models with Azure ML Azure ML and productivity with MLOps
Azure ML Learn how to use Azure Application Understand how to monitor data for Learn how to streamline and
Evaluate machine learning models for Insights to monitor a deployed Azure changing trends over time to ensure automate the end-to-end machine
fairness and mitigate predictive ML model. accuracy of model predictions. learning life cycle and tie it into
disparity in a machine learning existing DevOps processes.
model.

4 mins

IDC Report on Responsible AI

Understand how organizations
should approach Responsible AI from
a practitioner’s perspective.
Check out the optional resources to the right. Then use the navigation bar to return to your
learning journey or click ahead for week 4.
 Back to Learning Journey
WEEK 4

Week 4
Scale your skills with ML on Spark
Apache Spark offers another optimal way of creating ML models. Since it’s an
optimized engine for distributed processing, it can be very useful when creating
models with big data. Azure Databricks offers the best Spark experience on Azure.
With Azure Databricks you can access advanced machine learning capabilities,
combine data at any scale, and collaborate effectively on shared projects.

Spend your last week discovering the capabilities of Azure Databricks and its salient
features for ML on Spark. You’ll get tons of hands on experience- from performing
data transformations to building machine learning workflows to deploying and
serving models. There’s so much to explore here!

Click below to access the training or click ahead to view the training modules.

Click to begin training >

 Back to Learning Journey
WEEK 4

Perform data science with Azure Databricks Optional resources

Click the box to launch each module. Once completed, check the box to easily track your progress. Then use the To learn even more click on these
navigation bar to click ahead to complete your training. optional trainings.

55 mins 33 mins 1 hr 45 mins 3 hours

1. Describe Azure Databricks 2. Spark architecture 3. Read and write data in 4. Work with DataFrames in MLOps with MLFlow and Azure
fundamentals Azure Databricks Azure Databricks Databricks – Watch on demand
Discover the capabilities of Azure
Databricks and Apache Spark and Understand the architecture of an Perform data-handling functions with Execute data transformations in Watch the talks from our MLOps
how to identify tasks well suited for Virtual Event to learn best practices
Azure Databricks Spark Cluster and large datasets from multiple sources DataFrames and actions to display
for managing the ML lifecycle on
them. Spark Jobs. in different raw formats. transformed data. Databricks with MLflow.

35 mins 45 mins 1 hr 55 mins 6 mins read + sample notebooks

5. Work with user-defined 6. Build and query a Delta 7. Perform machine learning 8. Train a machine learning Running Tensorflow pipelines
functions Lake with Azure Databricks model with Azure Databricks
Learn how to write and perform ETL Use Delta Lake to create, append, Learn how to use PySpark’s ML Understand the building blocks in Get guidance on installing
(Extract, Transform, Load) functions and upsert data to Apache Spark package to build key components of the Spark’s ML library and build TensorFlow on Azure Databricks
and running TensorFlow programs.
with User-Defined Functions (UDFs). tables. the machine learning workflows. pipelines for common data
featurization tasks.

35 mins 33 mins 36 mins 25 mins 4 mins read + sample notebooks

9. Work with MLflow in Azure 10. Perform model selection 11. Deep learning with 12. Work with Azure Machine Distributed deep learning
Databrick with hyperparameter tuning Horovod for distributed Learning to deploy serving with Horovod
Track experiments, log metrics, and Know how to use modules from the training models Run distributed deep learning
compare runs. Plus, track experiment Spark’s ML library for See how Azure Databricks supports Register, package, and deploy a trained workloads on Azure Databricks
metrics, parameters, artifacts, and hyperparameter tuning and model enables distributed, deep learning model to Azure Container Instance and using Uber’s Horovod framework.
more. selection. training jobs on Spark. Azure Kubernetes Service.
 Back to Learning Journey

Congratulations on completing your journey! We hope seeing

the possibilities of Azure Machine Learning and Azure
Databricks has expanded your imagination. Now, you’re ready
to take on real-world ML projects. What’s next?

Schedule your certification exam

Show the world you’re equipped with the skills needed to implement an ML solution on Azure.

Engage with us on our resource page

Get the latest updates and best practices from the Product Management team.

Check out the AI for Developers journey

Try our developer centric AI services learning journey to create sophisticated apps with cognitive
capabilities.

Looking for more?

This is just the beginning of realizing what machine learning on Azure can do. Click ahead to
check out additional resources to keep learning and get started easily.
 Back to Learning Journey

Additional resources
Click the box to access the resource.

eBook eBook Paper Paper Infosheet

Mastering Azure Principles of Data Four real-life Machine Standardizing the ML MLOps Best Practices
Machine Learning Science Learning use cases Lifecycle Learn best practices for
extending DevOps practices to
Explore this free e-book from Get a comprehensive Dive into four practical end- Find out how to solve for ML
Machine Learning pipelines on
Packt for hands-on guidance, beginner's guide to statistical to-end machine-learning use lifecycle challenges compared
Azure.
real examples, and executable techniques and theory. cases on Azure Databricks. to the traditional software
code on Azure ML. development lifecycle.

Demo Demo Sample Code Sample Code Sample Code

Responsible ML Demo MLOps Demo NLP Recipe Recommenders Recipe Computer Vision
Watch this interactive demo Watch this interactive demo See how to leverage recent Get examples and best Recipe
which showcases model which showcases reproducible advances in NLP algorithms, practices for building
ML pipelines, deployment, neural architectures, and recommendation systems. See examples and best
interpretability, fairness and
CI/CD, and governance of ML distributed ML systems. practices for building
differential privacy in action.
projects. computer vision systems.