Predictive Modeling: Train your own machine-learning model & make predictions
      Back to home
      1. The AI & Analytics Engine Knowledge Hub
      2. Build from Scratch: Classification and Regression guide
      3. Predictive Modeling: Train your own machine-learning model & make predictions

      Create a classification or regression app

      In this article, you will learn how to create a machine-learning application for classification and regression tasks, to predict a variable/column from a dataset

      A classification or regression app on the AI & Analytics Engine forms a namespace for your feature sets, models, and deployments that is defined by the scope of a specific machine-learning task, which is to predict a selected variable/column from your chosen dataset.

      Note: For a more comprehensive definition, see What is an APP? 

      To get started click the “Build from scratch” button

      Within “Your first project” select “Build from scratch“ button.

      Clicking on build from scratch

      Step 1 - Define your ML problem

      1. Select the dataset ( The example dataset is german_credit_scrore - Processed) from the dropdown

      2. Select Predict a variable/column as the problem

      3. Select the column you want to predict (for the example dataset the column is SeriousDlqin2yrs)

      4. Click “Next”

      staging_app_1

       Step 2 - Select prediction type

      Based on the nature of the target column, the Engine will give you guidance by:

      • Suggesting the most suited prediction type from the list of available options,

      • Disabling the inappropriate options,

      To reduce the likelihood of failure in the next steps.

      In this step:

      1. Confirm if the best-suited option matches your problem, which is Answer a Yes/No question

        1. or select it from the enabled options

        2. or check for the reason why the desirable prediction type is disabled and fix the issue

      2. Because the selected prediction type is Answer a Yes/No question, define the Yes answer to the question by selecting 1 from the two available values (in the example dataset this would be for the column SeriousDlqin2yrs)

        1. the Engine automatically detects the Yes answer from your data

      3. Click “Next”

      For more information on why an option is deemed “most suited”, hover on the information icon of the option.

      If the prediction type you want to select is disabled, hover on the warning icon to view the reason and adjust your data accordingly before trying again.

      staging_app_2

      Step 3 - Review and run

      1. Give the app a name

      2. Review the prediction type

      The Engine automatically set the advanced configurations for apps that predict a variable. This article outlines the available advanced configurations for apps of this type. This article shows you how to configure the advanced configurations.

      Reviewing App

      NEXT: CREATE A FEATURE SET ⇒