Help with Lime Values (tabular explainer) for an existing model and dataframe

narquette Member Posts: 1
edited October 2022 in H2O-3 Open Source

Has anyone had any luck producing lime values (tabular explainer) for an existing model and dataframe with mixed data types (categorical and continuous)?

I've gone through tutorials and haven't had any luck getting values to be produced as the lime explainer expects data to be encoded and the model used data that wasn't encoded.

H2O Split Data and Model Information:
The model I'm using (GBM) was produced from AutoML with a dataset that wasn't encoded. The dataset used to trained the model has categorical values are in the original form and are listed as factor datatypes.

Lime Tabular Explainer Expectation:
The categorical data has to be encoded.

H2O Model Input
The data set being use to produce predictions expects the data format to match how it was trained (categorical features in the original form not encoded).