Is it possible to get an feature importance score for each levels of the categorical variable from BQ ML?
I have a random forecast classification model trained using BQ ML, where I have a categorical variable (string type) with multiple levels. I used ML.FEATURE_IMPORTANCE to pull the feature importance. However, for the categorical variable, the function only returns an overall importance score, rather than an importance score for each levels of the categorical variable like what I can get from Python. Can anyone help me get an importance score for each levels of the categorical variable from BQ ML? Thanks!