Microsoft Certified Azure Data Scientist Associate | DP 100 | Step By Step Activity Guides (Hands-On Labs)

This blog post covers Hands-On Labs that you must perform in order to learn Machine Learning, Data Science & clear the Azure Data Scientist Associate (DP-100) Certification.

This post helps you with your self-paced learning as well as for your team learning. There are 20 Hands-On Labs in this course.

Here’s the quick sneak-peak of how to start learning Data Science on Azure & to clear Azure Data Scientist Associate (DP-100) by doing Hands-on.

DP-100 learning path

Check our blog to know in more detail about the Azure Data Scientist Associate (DP-100) Certification

Activity Guides:

1.) Register For Azure Free Trial Account

The first thing required to perform the labs of DP-200 Implementing An Azure Data Scientist Exam is to get a Trial Account of Microsoft Azure. (You get 200 USD FREE Credit from Microsoft to practice)

Microsoft Azure is one of the top choices for any organization due to its freedom to build, manage, and deploy applications. In this activity guide, we will look at how to register for the Microsoft Azure FREE Trial Account.

You can Check out our blog to know more about how to create a Free Azure account.

Module 1: Getting Started With Azure Machine Learning

1) Creating An Azure Machine Learning Workspace

The workspace is the first resource to create for Azure Machine Learning, it provides a centralized place to work with all the assets like data, compute, model training code logged metrics, and trained models you create when you use Azure Machine Learning. The workspace keeps the history of all the training runs to pick the best model out of it.

azure ml workspace

2) Working With Azure Machine Learning Tools

In this guide, we will learn how to use the Azure Machine Learning studio interface, Azure Machine Learning SDK and developer tools like Visual Studio Code and Jupyter Notebooks to work with the assets in the workspace.

azure ml studio

Check out: Overview of Azure Machine Learning Service

Module 2: Visual Tools for Machine Learning.

1) Run An Automated Machine Learning Experiment

Azure Machine Learning includes an automated machine learning capability that leverages the scalability of cloud computing to automatically try multiple pre-processing techniques and model-training algorithms to find the best performing models for the data.

Automated ML

2.) Deploy And Test The Predictive Service (Automated ML)

After using automated machine learning to train some models, the best performing models can be deployed as a service for client applications to use.

3) Creating A Training Pipeline With The Azure ML Designer

A training pipeline has multiple tasks to perform like prepare data, train, deploy, and evaluate models. Then convert that training pipeline to a resulting pipeline that can be used to predict values for the new data.

azure ml pipeline

Read more: MLOps is based on DevOps principles and practices that increase the efficiency of workflows and improves the quality and consistency of the machine learning solutions.

4) Deploying A Service With The Azure ML Designer

Azure ML Designer is a Drag & Drop interface for creating machine learning models. It helps to visually connect datasets & modules

azure ml designer

Module 3: Running Experiments And Training Models

1) Running Experiments

Run is implementing a Python code that does a machine learning task, like training a model. In a run we can log metrics and upload results to Azure cloud, to keep track of experiments.

The experiment is a set of related runs. For example, if we train different models for solving a single problem, then we can group these runs under the same experiment, and afterward, we can compare their results.

running experiments in azure ml

2) Training And Registering Models

Training a Machine Learning model is a very simple process. Load the data on which the ML model needs to be created and then use a single line of code using a package like scikit-learn.

A Registered Model is a logical container of one or more files that creates your model. For example, if a model is stored in multiple files, we can register them as a single model in the workspace. We can register a model in three ways.

Using SDK
Using CLI
Visual Studio Code

Module 4: Working With Data

This module will cover how to create Datastores and Datasets in the Azure ML workspace.

1) Work With Data

Datastores are used to store connection information to various Azure storage services so we can access them by name and don’t need to remember the connection information.

A Dataset is a reference to the data in Datastore. By creating a dataset we are creating a reference to the data source location, and a copy of its metadata. Because the data remains in its source location, there is no extra storage cost. There are two types of Datasets.

Tabular Dataset
File Dataset

dataset in azure ML

Read more about Datastores and datasets in our blog at Working With Azure Datastores and Datasets.

Module 5: Working With Compute

1) Work With Compute

An Environment consists of Python packages, environment variables, and Docker settings that are used in machine learning experiments, including data preparation, training, and deployment of an ML model.

azure machine learning environment

We can train the model on various resources or the environment. These are called Compute Targets. A compute target can be a local machine or a cloud resource, such as an Azure Machine Learning Compute, Azure HDInsight, or a remote virtual machine.

compute target in azure ml

Module 6: Orchestrating Operations With Pipelines

1) Create And Publish A Pipeline

ML pipelines are used to create a workflow that integrates together various ML phases, and then publish that pipeline into your Azure ML workspace to access later or share with others.

We can publish a pipeline to run it with different inputs later. This way we don’t need to create different pipelines for different Datasets.

In this Hands-On we will learn how to create and publish an ML Pipeline using Azure Machine Learning SDK.

azure ml pipeline

Module 7: Deploying And Consuming Models

1) Creating A Real-time Inferencing Service

In this Hands-On guide, we will learn how to create a Real-time inferencing model using the inference pipeline. These inferences are based on a single observation of data at run time. The real-time inferencing model doesn’t require preprocessing.

2) Creating A Batch Inferencing Service

Batch Inferencing also called Offline Inferencing. It is the process of making predictions based on some batch of observations. The batch jobs are typically generated on some fixed schedule (e.g. hourly, daily).

batch inference in azure ml

Module 8: Training Optimal Models

1) Tuning Hyperparameters

Hyperparameters are adjustable parameters that we select to train a model. For example, to train a deep learning neural network we need to decide the number of hidden layers in the network and the number of nodes in each layer before we train the model. Selecting suitable hyperparameters is called Hyperparameters tuning.

Read more about it in our blog at Hyperparameter Tuning in Azure.

tuning hyperparameters

2) Using Automated Machine Learning

Automated machine learning is the process of automating the development of ML model and iterative tasks of machine learning model development. It allows data scientists, analysts, and developers to build ML models with high scale, efficiency, and productivity.

We use automated ML when we need to train & tune a model using the specified target metric.

automated machine learning

Module 9: Responsible Machine Learning

1) Explore Differential Policy

Differential privacy is a technique designed to preserve the privacy of individual data points by adding noise to the data. This lab focuses on how to install the SmartNoise SDK and with this create an analysis in which noise is added to the source data.

2) Interpreting Models

In this guide, we will cover how to predict which features in the data are taking part in making inferences and making an optimized ML model also which features are affecting the model prediction so we can remove these features from the dataset.

interpreting model

3.) Detect And Mitigate Unfairness

Machine learning models can incorporate unintentional bias, which can lead to issues with fairness. In this hands-on lab, we’ll use the Fairlearn package to analyze a model and explore disparity in prediction performance for different subsets of patients based on age.

Also Read: Our previous blog post on the Azure machine learning model.

Module 10: Monitoring Models

1) Monitoring A Model With Application Insights

Azure Machine Learning Logs monitoring is done using Azure Monitor. Azure Monitor provides a complete set of features to monitor Azure resources. We can analyze matrics for Azure machines using matrics from the Azure Monitor menu.

monitoring azure ml

2) Monitoring Data Drift

Data Drift is the change in the input data from the original data and this drift of Data affects the performance of the ML model created. This Hands-On guide covers how to monitor the Data Drift and how to deal with the change in the Data from the original Data.

To know more about Data Drift in Azure Machine Learning click here

data drift monitor

Related/References:

Next Task For You

To know more about the course, AI, ML, Data Science for beginners, why you should learn, Job opportunities, and what to study Including Hands-On labs you must perform to clear [DP-100] Microsoft Azure Data Scientist Associate Certification register for our FREE CLASS.

The post Microsoft Certified Azure Data Scientist Associate | DP 100 | Step By Step Activity Guides (Hands-On Labs) appeared first on Cloud Training Program.