Relevance Score
Introduction
From version 3.1, you can now check a score for the relevance between items on the screen for creating a new prediction model.
By selecting the item you want to predict, you can visually check the relevance of that item. This enables you to remove items with low relevance to the item you want to predict, or check for items that are too highly relevant.
This page describes how the relevance score can be utilized when creating a prediction model. For details on the method for calculating the relevance score, refer to “this page.”
The relevance score is calculated for numeric value items and string items when binary classification, multiclass classification, or numeric classification is selected. The relevance score is not calculated for text items or date/time items, or when using the time series prediction mode.