Skip to content

Create and edit the models

Insights Hub Quality Prediction Train Model Video videoplayer

This chapter describes the steps to create a prediction model,

To create a Predictive model, follow these steps:

  1. From the left navigation, click "Models". Prediction Models overview page is displayed.

    Quality_Prediction_Homepage

  2. Click create model to add a new prediction model.

  3. In "Select model information", perform the following actions:

    • Enter the "Model name" and the "Description of model version".
    • Select and configure the asset and its corresponding aspect with the process data used for model training and click “Next”.

    Select Model information

  4. In "Select input features", perform the following actions:

    • Select the time frame to obtain the IoT timeseries process data for training.
    • Click configure-window and select the identifier from the identifier selection window.
    • Check the box to use timestamp as an identifier.
    • Click add-feature and select the features from the feature selection window.
    • Enable the toggle to include the additional features "Time Based", "Count" and "Gap count".
    • Click "Apply".
    • The data chart is displayed according to the selected input features on the right-side panel as shown in the graphic and click “Next”.

    Select input features

  5. In "Filtering features", perform the following actions:

    • Click add-feature and add the process variables(features) from the IoT timeseries model from the feature selection window and Click "Select feature".
    • Enter the filter values to select data corresponding to the productive phases of the process (for example select relevant values of the process status or machine programs).
    • Check the box and enter the outlier value to remove outlier values during manufacturing phase.
    • Click "Apply" and click "Next".

    Filtering features

  6. In "Aggregation", perform the following actions:

    • Check the "Apply aggregation" check box to enable aggregation if multiple data points are collected during the production phase.
    • Select one or more combination of calculation methods (Minimum, Mean, Maximum and Standard Deviation) to formulate the aggregated variables of the manufacturing phase for each part number.
    • Select "Remove transients" check box and enter the value 1, if the first and the last data point of the production phase have to be excluded from the calculations.

    aggregation

    Info

    • "Remove transients" is selected if there are any artificial outliers in the ramp up and ramp down transient phases due to the delays in the data transfer from the physical machine.
    • "Aggregation" is not selected if the process data is already aggregated or only single datapoint data is collected during the manufacturing phase.

  7. In "Select quality data", perform the following actions:

    • Select the Quality Data file in csv format to train the model from Integrated Data Lake and review the file structure.
    • In "File schema configuration", select the feature type for the features (quality variables).
    • Click "Apply" and click “Next”.

    Select quality data

    Info

    The quality data file should already be uploaded to the selected folder in the Integrated Data Lake by the user. The file should contain defined measurement parameter (prediction target) representing product quality and required variables to be linked to the process parameters for building a process model. For information on feature and feature types, refer to Prerequisites.

  8. In “Select quality variables”, perform the following actions:

    • Select the process parameters for the quality variables. For information on quality variables, refer to Prerequisites.
    • Click "Apply" and click “Next”.

    Select quality variables

  9. In “Configure model”, select the model type and parameters in the user dialog window as shown below. The following are the regression model types available to be selected to train the model:

    • Linear Regression
    • Decision Tree
    • Random Forest
    • Gradient Boosting
    • XGBoost
    • Neural Network

Configure model

If “Optimized Parameters” is not selected, the user can enter the values of ML model parameters. Default model type and optimized parameters are recommended.
Click “View Training results” to start a model training. The model status changes to “In Progress” and the user is navigated to the "Training Board".

Last update: February 21, 2025