I will like to know if there is a handy tool like the simpleImputer in sklearn that helps with imputing missing data in tensorflow. I have only seen references to tfx, but i dont want to create a transform pipeline with apache beam yet, any help on this will be much appreciated. Thank you
One usually deals with such data issues at a early stage in the modelling process and it is separate from the model per se. Don’t you want to use sklearn’s simpleImputer you mentioned on your dataset and then plug these data objects/arrays of data to your TF model? But maybe I don’t understand what you are after.
I will like to have the imputation imbedded in the model, this will help deal with missing values in production.For example sklearn simple-imputer can be used to create a pipeline object that includes the preprocessing and prediction in a saved model object
I have the same question, and ideally in a way that can be converted to TfLite