10. Refit Model

In the previous tutorials, you successfully created and evaluated models.

In this tutorial, you’ll learn how to:

evaluate your models on a holdout set
refit a model using the most recent data — without needing a separate validation set. The refitted model automatically reuses the tuned parameters from the original model, including the number of trees determined by early stopping.

Step 1: Create Holdout Observation Table¶

Navigate to the Observation Table catalog under the Formulate section:
Select the Applications with Credit Default target table and go the About tab. Scroll down and click .
Create Applications Q1 2025 as follows:
- Name: "Applications Q1 2025"
- Description: "Credit Default Observations for holdout Q1 2025."
- Purpose: Validation-Test
- Sample Rows: 0
- Sampling Date Range: January 1, 2025 - April 1, 2025
Click .

Navigate to the Models Catalog under the Experiment section in the menu.

Select the top performing model of models generated by Ideation. Open Predict tab. click and select Applications Q1 2025 as the observation table.
Do the same for the model using the simplified Feature List.
Do the same for the model using the top SHAP importance Feature List.

Navigate to Leaderboard under the Experiment menu and configure the following:

Leaderboard configuration

Create Applications up to Dec 2024 training table as follows:
- Name: "Applications up to Dec 2024"
- Description: "Credit Default Observations for training up to Dec 2024."
- Purpose: Training
- Sample Rows: 0
- Sampling Date Range: April 1, 2019 - January 1, 2025
Select the model you want to refit. Open the Refit tab and click to start the refitting process.
Choose the Applications up to Dec 2024 observation table, then click to begin refitting.
When training completes, click on the refitted model and open predict tab to predict with the holdout data.
Navigate to the holdout Leaderboard to evaluate the refitted model.