I'm currently using MinMaxScaler() on my dataset. However, because my dataset is large I'm doing a first iteration pass in batches to compute the Min and Max Values for my Scaler. i'm using partial_fit() to help with this.

Anyway, for some of my features I do know their min and max values. Is there anyway I can explicity inform the scaler about these min and max values?

I'm currently using MinMaxScaler() on my dataset. However, because my dataset is large I'm doing a first iteration pass in batches to compute the Min and Max Values for my Scaler. i'm using partial_fit() to help with this.

Anyway, for some of my features I do know their min and max values. Is there anyway I can explicity inform the scaler about these min and max values?

• machine-learning
• scikit-learn
• scaling
• data-preprocessing

1 Answer 1

You could simply create your own function to transform your data:

def myMinMaxScaler(X, Xmin, Xmax):
    return (X - Xmin) / (Xmax - Xmin)

Another option could be to add rows (with the samples containing the min and max) at the end of your batches and after the transformation remove the added rows.