Missing Data Imputer#

Missing Data Imputer replaces missing continuous (denoted by NaN) or categorical values (denoted by special placeholder category such as '?') with a guess based on user-defined Strategy.

Interfaces: Transformer, Stateful, Persistable

Data Type Compatibility: Categorical and Continuous


# Name Default Type Description
1 continuous Mean Strategy The guessing strategy to employ for continuous feature columns.
2 categorical K Most Frequent Strategy The guessing strategy to employ for categorical feature columns.
3 categoricalPlaceholder '?' string The special placeholder category that denotes missing values.


use Rubix\ML\Transformers\MissingDataImputer;
use Rubix\ML\Other\Strategies\Percentile;
use Rubix\ML\Other\Strategies\Prior;

$transformer = new MissingDataImputer(new Percentile(0.55), new Prior(), '?');

Additional Methods#

This transformer does not have any additional methods.