Hello everyone.
I am using LIME to set up explanations about the decisions of different models with tabular data. For the training of AI models, I applied StandardScaler to the data in order to avoid overfitting, the problem is that this changes the scale of the explanations and the ranges that are next to the variables become useless (eg: 'severe_decelerations <= -0.27' says nothing because the data is standardized). Is there anything I can do about it (besides training again the models without the StandardScaler)? Sorry if I'm missing something obvious and thanks for the help.
Hello everyone. I am using LIME to set up explanations about the decisions of different models with tabular data. For the training of AI models, I applied StandardScaler to the data in order to avoid overfitting, the problem is that this changes the scale of the explanations and the ranges that are next to the variables become useless (eg: 'severe_decelerations <= -0.27' says nothing because the data is standardized). Is there anything I can do about it (besides training again the models without the StandardScaler)? Sorry if I'm missing something obvious and thanks for the help.