conditional_algorithm_parameters

Description

This configuration function allows for the specification of algorithm parameters that override global settings for specific subgroups within the dataset. These subgroups are identified based on the values in discriminant columns. By setting conditional parameters, users can tailor the algorithm's behavior to better fit the distinct characteristics or needs of different segments within the data. This targeted approach can significantly enhance model performance by optimizing parameters for specific conditions or scenarios within the dataset.

Steps impacted

training

Example

conditional_algorithm_parameters:
    group1:
        where:
            country: us
        alpha: 0.666
    group2:
        where:
           country: fr
        alpha: 0.777