Nice blog avi bro. I have a small doubt that what should we pass as group parameter? Also instead of group shuffle split, can we divide the whole dataset into train test and valid and make pipelines for avoiding the data leakage ( I'm saying by considering the tabular data) ? Will that technique works well for image data??
i think stratify is also similiar right? but does group split make sure the distribution uniformity?
Nice blog avi bro. I have a small doubt that what should we pass as group parameter? Also instead of group shuffle split, can we divide the whole dataset into train test and valid and make pipelines for avoiding the data leakage ( I'm saying by considering the tabular data) ? Will that technique works well for image data??