1.2 KiB
1.2 KiB
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
Parameters
| # | 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. |
Example
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.