- Article
- 13 minutes to read
Learn how to train a classification model with no-code AutoML using Azure Machine Learning automated ML in the Azure Machine Learning studio. This classification model predicts if a client will subscribe to a fixed term deposit with a financial institution.
With automated ML, you can automate away time intensive tasks. Automated machine learning rapidly iterates over many combinations of algorithms and hyperparameters to help you find the best model based on a success metric of your choosing.
You won't write any code in this tutorial, you'll use the studio interface to perform training. You'll learn how to do the following tasks:
- Create an Azure Machine Learning workspace.
- Run an automated machine learning experiment.
- Explore model details.
- Deploy the recommended model.
Also try automated machine learning for these other model types:
- For a no-code example of forecasting, see .
- For a code first example of an object detection model, see the Tutorial: Train an object detection model with AutoML and Python,
Prerequisites
An Azure subscription. If you don't have an Azure subscription, create a free account.
Download the bankmarketing_train.csv data file. The y column indicates if a customer subscribed to a fixed term deposit, which is later identified as the target column for predictions in this tutorial.
Create a workspace
An Azure Machine Learning workspace is a foundational resource in the cloud that you use to experiment, train, and deploy machine learning models. It ties your Azure subscription and resource group to an easily consumed object in the service.
There are many ways to create a workspace. In this tutorial, you create a workspace via the Azure portal, a web-based console for managing your Azure resources.
Sign in to the Azure portal by using the credentials for your Azure subscription.
In the upper-left corner of the Azure portal, select the three bars, then + Create a resource.
Use the search bar to find Azure Machine Learning.
Select Azure Machine Learning.
In the Machine Learning pane, select Create to begin.
Provide the following information to configure your new workspace:
Field Description Workspace name Enter a unique name that identifies your workspace. In this example, we use docs-ws. Names must be unique across the resource group. Use a name that's easy to recall and to differentiate from workspaces created by others. Subscription Select the Azure subscription that you want to use. Resource group Use an existing resource group in your subscription, or enter a name to create a new resource group. A resource group holds related resources for an Azure solution. In this example, we use docs-aml. Region Select the location closest to your users and the data resources to create your workspace. Storage account A storage account is used as the default datastore for the workspace. You may create a new Azure Storage resource or select an existing one in your subscription. Key vault A key vault is used to store secrets and other sensitive information that is needed by the workspace. You may create a new Azure Key Vault resource or select an existing one in your subscription. Application insights The workspace uses Azure Application Insights to store monitoring information about your deployed models. You may create a new Azure Application Insights resource or select an existing one in your subscription. Container registry A container registry is used to register docker images used in training and deployments. You may choose to create a resource or select an existing one in your subscription. After you're finished configuring the workspace, select Review + Create.
Select Create to create the workspace.
Warning
It can take several minutes to create your workspace in the cloud.
When the process is finished, a deployment success message appears.
To view the new workspace, select Go to resource.
From the portal view of your workspace, select Launch studio to go to the Azure Machine Learning studio.
(Video) Azure Machine Learning Studio Tutorial
Important
Take note of your workspace and subscription. You'll need these to ensure you create your experiment in the right place.
Sign in to the studio
You complete the following experiment set-up and run steps via the Azure Machine Learning studio at https://ml.azure.com, a consolidated web interface that includes machine learning tools to perform data science scenarios for data science practitioners of all skill levels. The studio is not supported on Internet Explorer browsers.
Sign in to Azure Machine Learning studio.
Select your subscription and the workspace you created.
Select Get started.
In the left pane, select Automated ML under the Author section.
Since this is your first automated ML experiment, you'll see an empty list and links to documentation.
Select +New automated ML job.
Create and load dataset
Before you configure your experiment, upload your data file to your workspace in the form of an Azure Machine Learning dataset. Doing so, allows you to ensure that your data is formatted appropriately for your experiment.
Create a new dataset by selecting From local files from the +Create dataset drop-down.
On the Basic info form, give your dataset a name and provide an optional description. The automated ML interface currently only supports TabularDatasets, so the dataset type should default to Tabular.
Select Next on the bottom left
On the Datastore and file selection form, select the default datastore that was automatically set up during your workspace creation, workspaceblobstore (Azure Blob Storage). This is where you'll upload your data file to make it available to your workspace.
Select Upload files from the Upload drop-down.
Choose the bankmarketing_train.csv file on your local computer. This is the file you downloaded as a prerequisite.
Select Next on the bottom left, to upload it to the default container that was automatically set up during your workspace creation.
When the upload is complete, the Settings and preview form is pre-populated based on the file type.
Verify that the Settings and preview form is populated as follows and select Next.
Field Description Value for tutorial File format Defines the layout and type of data stored in a file. Delimited Delimiter One or more characters for specifying the boundary between separate, independent regions in plain text or other data streams. Comma Encoding Identifies what bit to character schema table to use to read your dataset. UTF-8 Column headers Indicates how the headers of the dataset, if any, will be treated. All files have same headers Skip rows Indicates how many, if any, rows are skipped in the dataset. None The Schema form allows for further configuration of your data for this experiment. For this example, select the toggle switch for the day_of_week, so as to not include it. Select Next.
On the Confirm details form, verify the information matches what was previously populated on the Basic info, Datastore and file selection and Settings and preview forms.
Select Create to complete the creation of your dataset.
Select your dataset once it appears in the list.
Review the Data preview to ensure you didn't include day_of_week then, select Close.
Select Next.
(Video) Introduction to no code Azure Machine Learning
Configure job
After you load and configure your data, you can set up your experiment. This setup includes experiment design tasks such as, selecting the size of your compute environment and specifying what column you want to predict.
Select the Create new radio button.
Populate the Configure Job form as follows:
Enter this experiment name:
my-1st-automl-experimentSelect y as the target column, what you want to predict. This column indicates whether the client subscribed to a term deposit or not.
Select compute cluster as your compute type.
+New to configure your compute target. A compute target is a local or cloud-based resource environment used to run your training script or host your service deployment. For this experiment, we use a cloud-based compute.
Populate the Select virtual machine form to set up your compute.
Field Description Value for tutorial Location Your region that you'd like to run the machine from West US 2 Virtualmachinetier Select what priority your experiment should have Dedicated Virtualmachinetype Select the virtual machine type for your compute. CPU (Central Processing Unit) Virtualmachinesize Select the virtual machine size for your compute. A list of recommended sizes is provided based on your data and experiment type. Standard_DS12_V2 Select Next to populate the Configure settings form.
Field Description Value for tutorial Compute name A unique name that identifies your compute context. automl-compute Min / Max nodes To profile data, you must specify 1 or more nodes. Min nodes: 1
Max nodes: 6Idle seconds before scale down Idle time before the cluster is automatically scaled down to the minimum node count. 120 (default) Advanced settings Settings to configure and authorize a virtual network for your experiment. None Select Create to create your compute target.
This takes a couple minutes to complete.
After creation, select your new compute target from the drop-down list.
Select Next.
On the Select task and settings form, complete the setup for your automated ML experiment by specifying the machine learning task type and configuration settings.
Select Classification as the machine learning task type.
Select View additional configuration settings and populate the fields as follows. These settings are to better control the training job. Otherwise, defaults are applied based on experiment selection and data.
Additionalconfigurations Description Valuefortutorial Primary metric Evaluation metric that the machine learning algorithm will be measured by. AUC_weighted Explain best model Automatically shows explainability on the best model created by automated ML. Enable Blocked algorithms Algorithms you want to exclude from the training job None Additionalclassification settings These settings help improve the accuracy of your model Positive class label: None Exit criterion If a criteria is met, the training job is stopped. Trainingjobtime (hours): 1
Metricscorethreshold: NoneConcurrency The maximum number of parallel iterations executed per iteration Maxconcurrentiterations: 5 Select Save.
Select Next.
On the [Optional] Validate and test form,
- Select k-fold cross-validation as your Validation type.
- Select 2 as your Number of cross validations.
Select Finish to run the experiment. The Job Detail screen opens with the Job status at the top as the experiment preparation begins. This status updates as the experiment progresses. Notifications also appear in the top right corner of the studio to inform you of the status of your experiment.
Important
Preparation takes 10-15 minutes to prepare the experiment run.Once running, it takes 2-3 minutes more for each iteration.
In production, you'd likely walk away for a bit. But for this tutorial, we suggest you start exploring the tested algorithms on the Models tab as they complete while the others are still running.
Explore models
Navigate to the Models tab to see the algorithms (models) tested. By default, the models are ordered by metric score as they complete. For this tutorial, the model that scores the highest based on the chosen AUC_weighted metric is at the top of the list.
While you wait for all of the experiment models to finish, select the Algorithm name of a completed model to explore its performance details.
The following navigates through the Details and the Metrics tabs to view the selected model's properties, metrics, and performance charts.
Model explanations
While you wait for the models to complete, you can also take a look at model explanations and see which data features (raw or engineered) influenced a particular model's predictions.
These model explanations can be generated on demand, and are summarized in the model explanations dashboard that's part of the Explanations (preview) tab.
To generate model explanations,
Select Job 1 at the top to navigate back to the Models screen.
Select the Models tab.
For this tutorial, select the first MaxAbsScaler, LightGBM model.
Select the Explain model button at the top. On the right, the Explain model pane appears.
Select the automl-compute that you created previously. This compute cluster initiates a child job to generate the model explanations.
Select Create at the bottom. A green success message appears towards the top of your screen.
Note
The explainability job takes about 2-5 minutes to complete.
Select the Explanations (preview) button. This tab populates once the explainability run completes.
On the left hand side, expand the pane and select the row that says raw under Features.
Select the Aggregate feature importance tab on the right. This chart shows which data features influenced the predictions of the selected model.
In this example, the duration appears to have the most influence on the predictions of this model.
Deploy the best model
The automated machine learning interface allows you to deploy the best model as a web service in a few steps. Deployment is the integration of the model so it can predict on new data and identify potential areas of opportunity.
For this experiment, deployment to a web service means that the financial institution now has an iterative and scalable web solution for identifying potential fixed term deposit customers.
Check to see if your experiment run is complete. To do so, navigate back to the parent job page by selecting Job 1 at the top of your screen. A Completed status is shown on the top left of the screen.
Once the experiment run is complete, the Details page is populated with a Best model summary section. In this experiment context, VotingEnsemble is considered the best model, based on the AUC_weighted metric.
We deploy this model, but be advised, deployment takes about 20 minutes to complete. The deployment process entails several steps including registering the model, generating resources, and configuring them for the web service.
Select VotingEnsemble to open the model-specific page.
Select the Deploy menu in the top-left and select Deploy to web service.
(Video) Azure Machine Learning | Auto ML | Create ML Model with ML Studio and AutoML | Step by Step GuidePopulate the Deploy a model pane as follows:
Field Value Deployment name my-automl-deploy Deployment description My first automated machine learning experiment deployment Compute type Select Azure Container Instance (ACI) Enable authentication Disable. Use custom deployments Disable. Allows for the default driver file (scoring script) and environment file to be auto-generated. For this example, we use the defaults provided in the Advanced menu.
Select Deploy.
A green success message appears at the top of the Job screen, and in the Model summary pane, a status message appears under Deploy status. Select Refresh periodically to check the deployment status.
Now you have an operational web service to generate predictions.
Proceed to the Next Steps to learn more about how to consume your new web service, and test your predictions using Power BI's built in Azure Machine Learning support.
Clean up resources
Deployment files are larger than data and experiment files, so they cost more to store. Delete only the deployment files to minimize costs to your account, or if you want to keep your workspace and experiment files. Otherwise, delete the entire resource group, if you don't plan to use any of the files.
Delete the deployment instance
Delete just the deployment instance from Azure Machine Learning at https://ml.azure.com/, if you want to keep the resource group and workspace for other tutorials and exploration.
Go to Azure Machine Learning. Navigate to your workspace and on the left under the Assets pane, select Endpoints.
Select the deployment you want to delete and select Delete.
Select Proceed.
Delete the resource group
Important
The resources that you created can be used as prerequisites to other Azure Machine Learning tutorials and how-to articles.
If you don't plan to use any of the resources that you created, delete them so you don't incur any charges:
In the Azure portal, select Resource groups on the far left.
From the list, select the resource group that you created.
Select Delete resource group.
Enter the resource group name. Then select Delete.
Next steps
In this automated machine learning tutorial, you used Azure Machine Learning's automated ML interface to create and deploy a classification model. See these articles for more information and next steps:
Consume a web service
- Learn more about automated machine learning.
- For more information on classification metrics and charts, see the Understand automated machine learning results article.
- Learn more about featurization.
- Learn more about data profiling.
Note
This Bank Marketing dataset is made available under the Creative Commons (CCO: Public Domain) License. Any rights in individual contents of the database are licensed under the Database Contents License and available on Kaggle. This dataset was originally available within the UCI Machine Learning Database.
[Moro et al., 2014] S. Moro, P. Cortez and P. Rita. A Data-Driven Approach to Predict the Success of Bank Telemarketing. Decision Support Systems, Elsevier, 62:22-31, June 2014.
FAQs
How do you train ML models on Azure? ›
- Connect to your Azure Machine Learning workspace.
- Create your compute resource and job environment.
- Create your training script.
- Create and run your command job to run the training script on the compute resource, configured with the appropriate job environment.
February 27, 2023. Databricks AutoML helps you automatically apply machine learning to a dataset. You provide the dataset and identify the prediction target, while AutoML prepares the dataset for model training. AutoML then performs and records a set of trials that creates, tunes, and evaluates multiple models.
What are the key steps in the AutoML process? ›There are generally eight steps in the AutoML process: data ingestion, data preparation, data engineering, model selection, model training, hyperparameter tuning, model deployment, and model updates.
Which steps can automate by AutoML? ›- Raw data processing.
- Feature engineering and feature selection.
- Model selection.
- Hyperparameter optimization and parameter optimization.
You can use AutoML to build on Google's machine learning capabilities to create your own custom machine learning models that are tailored to your business needs, and then integrate those models into your applications and web sites.
How do you implement MLOps in Azure? ›- Deploy and scale containers on managed Kubernetes.
- Deploy and scale containers on managed Red Hat OpenShift.
- Azure Container Apps. ...
- Execute event-driven serverless code functions with an end-to-end development experience.
- Run containerized web apps on Windows and Linux.
- Azure Container Instances.
- Develop and create a model in a training environment. To deploy a machine learning application, you first need to build your model. ...
- Optimize and test code, then clean and test again. ...
- Prepare for container deployment. ...
- Plan for continuous monitoring and maintenance.
Generally speaking, the rule of thumb regarding machine learning is that you need at least ten times as many rows (data points) as there are features (columns) in your dataset. This means that if your dataset has 10 columns (i.e., features), you should have at least 100 rows for optimal results.
How much does it cost to train an ML model? ›From the above results, you can see that an AWS cloud instance called c6g. 8xlarge. od will take 6.19 hours to train the machine learning model at a total cost of $6.75. The cost may be high compared to other cloud instances, but you will save a lot of time if you plan to run multiple machine learning experiments.
How many pictures does it take to train a ML model? ›the classes are trained with many images. Usually around 100 images are sufficient to train a class. If the images in a class are very similar, fewer images might be sufficient. the training images are representative of the variation typically found within the class.
What is the disadvantage of AutoML? ›
The main criticisms of AutoML solutions are: 1 Control - Can't alter generated solutions. 2 It doesn't do enough - Most of the work is elsewhere. 3 Quality of results - Users don't want to be held back.
Which algorithm is used by AutoML? ›AutoML Systems
AutoWEKA is an approach for the simultaneous selection of a machine learning algorithm and its hyperparameters; combined with the WEKA package it automatically yields good models for a wide variety of data sets.
- Data Collection. → The quantity & quality of your data dictate how accurate our model is. ...
- Data Preparation. → Wrangle data and prepare it for training. ...
- Choose a Model. ...
- Train the Model. ...
- Evaluate the Model. ...
- Parameter Tuning. ...
- Make Predictions.
- Stage 1: Collect and prepare data. ...
- Stage 2: Make sense of data. ...
- Stage 3: Use data to answer questions. ...
- Stage 4: Create predictive applications.
AutoML is primarily used for supervised learning applications, i.e., regression and classification. Automl techniques can also be used for unsupervised learning and reinforcement learning.
Will AutoML replace machine learning? ›2. AutoML will not replace most data science professions; rather, it will assist experts in completing their assignments more quickly.
Which AutoML is best? ›- DataRobot. ...
- MLBox. ...
- Auto Sklearn. ...
- TPOT. ...
- H2O. ...
- Auto Keras. ...
- Google Cloud AutoML. Cloud AutoML uses a neural network architecture. ...
- Uber Ludwig. The goal of the Uber Ludwig project is to automate modern deep learning systems with a minimal amount of code.
AutoML objectives and benefits overlap with those of MLOps — a broader discipline with focus not only on automation but also on cross-functional collaboration within machine learning projects.
What are the main 3 types of ML models? ›Amazon ML supports three types of ML models: binary classification, multiclass classification, and regression. The type of model you should choose depends on the type of target that you want to predict.
What is the difference between AutoML and custom model? ›You create a custom model by training it using a prepared dataset. AutoML Translation uses the items from the dataset to train the model, test it, and evaluate its performance. You review the results, adjust the training dataset as needed and train a new model using the improved dataset.
What is the difference between AutoML and Vertex AI? ›
Vertex AI combines data engineering, data science, and ML engineering workflows, enabling your teams to collaborate using a common toolset. Vertex AI provides several options for model training: AutoML allows you to train tabular, image, text, or video data without writing code or preparing data splits.
What is Azure AutoML? ›Azure AutoML is a cloud-based service that can be used to automate building machine learning pipelines for classification, regression and forecasting tasks. Its goal is not only to tune hyper-parameters of a given model, but also to identify which model to use and how to pre-process the input dataset.
What is Azure DevOps vs Azure MLOps? ›One key difference between MLOps and DevOps is that MLOps places a greater emphasis on automated machine learning tasks, such as training models. DevOps, on the other hand, focuses more on traditional software development tasks such as code builds and deployments.
Does Azure ML use MLflow? ›Azure Machine Learning workspaces are MLflow-compatible, which means you can use MLflow to track runs, metrics, parameters, and artifacts with your Azure Machine Learning workspaces.
What is the best way to train a model in machine learning? ›- Step 1: Begin with existing data. Machine learning requires us to have existing data—not the data our application will use when we run it, but data to learn from. ...
- Step 2: Analyze data to identify patterns. ...
- Step 3: Make predictions.
Continuous training is an aspect of machine learning operations that automatically and continuously retrains machine learning models to adapt to changes in the data before it is redeployed. The trigger for a re-build can be data change, model change, or code change.
What is the 10 times rule machine learning? ›The most common way to define whether a data set is sufficient is to apply a 10 times rule. This rule means that the amount of input data (i.e., the number of examples) should be ten times more than the number of degrees of freedom a model has. Usually, degrees of freedom mean parameters in your data set.
Can you train a model with multiple datasets? ›Multiple Dataset feature allows you to train your model on multiple datasets which helps fine-tune your model to offer accurate recommendations and improve the end-user experience over time.
Is 1000 data enough for machine learning? ›Because then you can think about which data you need and what you have to do for it. As a rule, however, you need far more than 50 observations. Our experience shows that everything over 1,000 goes in the right direction. But we have also seen problems where even 1,000,000 data points were just not enough.
Why do ML models fail? ›Machine learning model training that doesn't generalize
With a clearly defined business problem and targeted success metrics, your potential pitfalls get more technical. During the model training stage, issues related to your training data or model fit are the likeliest culprit for future failure.
How long is AutoML training? ›
AutoML Natural Language uses early stopping to produce the best possible model without overfitting. For classification models, the average training time is around 6 hours, with a maximum of 24 hours.
Can you sell a machine learning model? ›Selling a Model
Building and training your own models allows you to sell them in the marketplace after a review process. This example assumes you've created a model using the procedure outlined in the Building a Model and Deploying a Model tutorials.
Convolutional Neural Networks (CNNs) is the most popular neural network model being used for image classification problem.
What is a good sample size for machine learning? ›If you've talked with me about starting a machine learning project, you've probably heard me quote the rule of thumb that we need at least 1,000 samples per class.
How big of a data set do you need to train AI? ›Estimate using the rule of 10: For an initial estimation for the amount of data required, you can apply the rule of 10, which recommends that the amount of training data you need is 10 times the number of parameters – or degrees of freedom – in the model.
What are the challenges in AutoML? ›In conclusion, Auto ML implementation is raising a few challenges such as parallelization, result collection, resource optimization, iteration, etc. Machine learning pipelines provide a solution to answer those challenges with a clear definition of the process and automation features.
What are the limits of AutoML? ›AutoML Tables enforces the following limits on training data: Maximum size of 100 GB. Between 1,000 and 200,000,000 rows. Between 2 and 1,000 columns.
Does AutoML clean data? ›Data Cleaning for AutoML.
Some existing AutoML frameworks already include data cleaning mechanisms. For example, AutoGluon is a Python library for AutoML with tabular data [5]. It automatically performs MV imputation and outlier detection before the ML pipeline in both a model-agnostic and model-specific way.
One of the best sources for classification datasets is the UCI Machine Learning Repository. The Mushroom dataset is a classic, the perfect data source for logistic regression, decision tree, or random forest classification practice.
Which is the easiest ML algorithm? ›K-means clustering
K-means clustering is one of the simplest and a very popular unsupervised machine learning algorithms.
What is the simplest classification algorithm? ›
k-Nearest Neighbors
kNN stands for “k-nearest neighbor” and is one of the simplest classification algorithms. The algorithm assigns objects to the class that most of its nearest neighbors in the multidimensional feature space belong to.
- Open the AutoML Translation UI. ...
- Select the dataset you want to use to train the custom model. ...
- When you are done reviewing the dataset, click the Train tab just below the title bar.
- Click Start Training. ...
- Specify a name for the model.
- Click Start Training to begin training your custom model.
- In the Test Plans web portal, open the test plan and select a test suite that contains the automated tests.
- Select the test(s) you want to run, open the Run menu, and choose Run test. ...
- Choose OK to start the testing process.
Microsoft combined three unique services—Azure Monitor, Log Analytics, and Application Insights—under the umbrella of Azure Monitor to provide powerful end-to-end monitoring of your applications and the components they rely on. Log Analytics and Application Insights are now features of Azure Monitor.
What is the difference between Azure monitor and application Insights? ›Application Insights is an extension of Azure Monitor and provides Application Performance Monitoring (also known as “APM”) features. APM tools are useful to monitor applications from development, through test, and into production in the following ways: Proactively understand how an application is performing.
What are the 7 things Azure security monitor can do? ›- Applications.
- Virtual machines.
- Guest operating systems.
- Containers.
- Databases.
- Security events in combination with Azure Sentinel.
- Networking events and health in combination with Network Watcher.
AutoML Systems
AutoWEKA is an approach for the simultaneous selection of a machine learning algorithm and its hyperparameters; combined with the WEKA package it automatically yields good models for a wide variety of data sets.
- Sign in to the Azure portal.
- Search for and select Automation Accounts.
- On the Automation Accounts page, select your Automation account from the list.
- From the Automation account, select Runbooks under Process Automation to open the list of runbooks.
- Click Create a runbook.
The currently supported versions are: Python 2.7 (GA), Python 3.8 (preview), and Python 3.10 (preview). Graphical runbook based on Windows PowerShell and created and edited completely in the graphical editor in Azure portal.
How do you write automated test cases in Azure DevOps? ›- Step 2: Create a new Project.
- Step 3: Create a Pipeline.
- Step 4: Link your code repo.
- Step 5: Select the task.
- Step 6: Configure Maven goals.
- Step 7: Pipeline run.