I was wondering what's the best way to preprocess new samples for my ML classifier.
I have a raw data with about 3000 samples. I'm preprocessing it with some LabelEncoders and TargetEncdoders for categorical variables and StandardScalers for numerical values.
What's the best way to preprocess a single sample for prediction, the same way as raw_data was. How to store parameters, calculated by my encoders.
EX. I have categorical variable, which is transformed with LabelEncoder and than TargetEncoder, and category "A" is transformed to value 0.5 . I want to make sure my single sample, with category "A" will be transofrmed to the same numerical value.
Is it possible to do with Sklearn's pipelines?