Module imputer 

Simple imputer: fill missing (NaN) values per feature column.

SimpleImputer supports four imputation strategies:

• ImputeStrategy::Mean — replace NaN with the column mean
• ImputeStrategy::Median — replace NaN with the column median
• ImputeStrategy::MostFrequent — replace NaN with the most common value
• ImputeStrategy::Constant — replace NaN with a fixed constant value

Fitting ignores NaN values when computing statistics (e.g. the mean is the mean of all non-NaN values in that column). Columns that are entirely NaN at fit time are filled with F::zero() under Mean/Median and with the most frequent non-NaN value (defaulting to F::zero()) under MostFrequent.

Structs

FittedSimpleImputer

A fitted simple imputer holding one fill value per feature column.

SimpleImputer

An unfitted simple imputer.

Enums

ImputeStrategy

The strategy used to compute the fill value for each column.