Struct RareLabelEncoder 

pub struct RareLabelEncoder {
    pub columns: Vec<String>,
    pub threshold: f64,
    pub mapping: HashMap<String, HashMap<String, String>>,
    /* private fields */
}

Groups infrequent categories into a single “rare” label.

Fields

columns: Vec<String>

threshold: f64

mapping: HashMap<String, HashMap<String, String>>

Mapping from column to (category -> encoded label)

Implementations

impl RareLabelEncoder

pub fn new(columns: Vec<String>, threshold: f64) -> Self

Create a new RareLabelEncoder for the specified columns and threshold.

pub async fn fit(&mut self, df: &DataFrame) -> FeatureFactoryResult<()>

Fit computes frequencies and marks infrequent categories as “rare”.

pub fn transform(&self, df: DataFrame) -> FeatureFactoryResult<DataFrame>

Transform replaces each category with its encoded label.

Trait Implementations

impl Transformer for RareLabelEncoder

fn fit<'life0, 'life1, 'async_trait>( &'life0 mut self, df: &'life1 DataFrame, ) -> Pin<Box<dyn Future<Output = FeatureFactoryResult<()>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait,

Fit the transformer given a DataFrame.

fn transform(&self, df: DataFrame) -> FeatureFactoryResult<DataFrame>

Transform the input DataFrame, returning a new DataFrame with the transformation applied.

fn is_stateful(&self) -> bool

Returns true if the transformer is stateful (i.e. requires a call to fit before transform can be called).