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

Train machine learning model(s)

Train your own machine learning model(s)

In this article, you will train 2 machine learning models in your newly created app. 

Typically, Data Scientists will train multiple machine learning models to compare and select the best performing model.

1. Train New Model

In the app page, you will be able to select the TRAIN NEW MODEL option.

Step 1: Select a feature set

  • Feature Set: Select the sample feature set that we created in the create a feature set tutorial. Or proceed with the default feature set if you skipped the feature set tutorial.
  • Dataset version: Use the default V0. We will not bother with the Dataset version as it relates to the Continuous Learning function.

Click NEXT.

Step 2: Choose a model

Here is where we can choose the model/algorithm that we want. To assist with the selection process, the Engine's model recommender has provided estimations, as seen next to each model, the:

  • Predictive performance
  • Prediction Time
  • Training Time

Let's select two models for training:

  1. LightGBM Classifier
  2. XGBoost Classifier

Step 3: Configure

Like tuning a car, models can similarly be tuned for optimum performances. Here, we will rely on the default configurations provided by the Data Scientists at PI.EXCHANGE.

Click 'NEXT'. 

Note: For more information on configurations, see Advanced configurations for model training.

Step 4: Train Your Model

Here, we do a final review of the models' configurations prior to training.

Click on TRAIN MODEL for your models to begin training.