Temporal Features

class temporal_features.CyclicalFeaturesEncoder(columns: str | List[str], max_values: int | List[int])[source]

Encodes cyclical features using sine and cosine transformations.

fit(X: DataFrame, y: Series | None = None) → CyclicalFeaturesEncoder[source]

Fit method does nothing as no fitting is required.

Parameters:

X (pd.DataFrame) – Input DataFrame.
y (pd.Series, optional) – Target variable (ignored).

Returns:

Returns self.

Return type:

CyclicalFeaturesEncoder

transform(X: DataFrame) → DataFrame[source]

Encodes cyclical features.

Parameters:: X (pd.DataFrame) – Input DataFrame.
Returns:: DataFrame with new sine and cosine encoded columns.
Return type:: pd.DataFrame

class temporal_features.DataResampler(datetime_column: str, rule: str, aggregation_methods: str | Dict[str, str] = 'sum')[source]

Resamples the DataFrame based on a given frequency and aggregation method.

fit(X: DataFrame, y: Series | None = None) → DataResampler[source]

Fit method does nothing as no fitting is required.

Parameters:

X (pd.DataFrame) – Input DataFrame.
y (pd.Series, optional) – Target variable (ignored).

Returns:

Returns self.

Return type:

DataResampler

transform(X: DataFrame) → DataFrame[source]

Resamples the DataFrame.

Parameters:: X (pd.DataFrame) – Input DataFrame.
Returns:: Resampled DataFrame with aggregated values.
Return type:: pd.DataFrame

class temporal_features.DatePartExtractor(column: str, parts: List[str] | None = None, prefix: str | None = None)[source]

Extracts date parts from datetime columns.

fit(X: DataFrame, y: Series | None = None) → DatePartExtractor[source]

Fit method does nothing as no fitting is required.

Parameters:

X (pd.DataFrame) – Input DataFrame.
y (pd.Series, optional) – Target variable (ignored).

Returns:

Returns self.

Return type:

DatePartExtractor

transform(X: DataFrame) → DataFrame[source]

Extracts specified date parts from the datetime column.

Parameters:: X (pd.DataFrame) – Input DataFrame.
Returns:: DataFrame with extracted date parts.
Return type:: pd.DataFrame

class temporal_features.DatetimeConverter(columns: str | List[str], format: str | None = None, errors: str = 'raise')[source]

Converts specified columns to datetime format.

fit(X: DataFrame, y: Series | None = None) → DatetimeConverter[source]

Fit method does nothing as no fitting is required.

Parameters:

X (pd.DataFrame) – Input DataFrame.
y (pd.Series, optional) – Target variable (ignored).

Returns:

Returns self.

Return type:

DatetimeConverter

transform(X: DataFrame) → DataFrame[source]

Converts specified columns to datetime.

Parameters:: X (pd.DataFrame) – Input DataFrame.
Returns:: DataFrame with specified columns converted to datetime.
Return type:: pd.DataFrame

class temporal_features.LagFeatureCreator(columns: str | List[str], lags: List[int])[source]

Creates lag features for specified columns.

fit(X: DataFrame, y: Series | None = None) → LagFeatureCreator[source]

Fit method does nothing as no fitting is required.

Parameters:

X (pd.DataFrame) – Input DataFrame.
y (pd.Series, optional) – Target variable (ignored).

Returns:

Returns self.

Return type:

LagFeatureCreator

transform(X: DataFrame) → DataFrame[source]

Creates lag features.

Parameters:: X (pd.DataFrame) – Input DataFrame.
Returns:: DataFrame with new lag feature columns.
Return type:: pd.DataFrame

class temporal_features.RollingFeatureCreator(columns: str | List[str], window_size: int, statistics: List[str] = ['mean'])[source]

Creates rolling statistics for specified columns.

fit(X: DataFrame, y: Series | None = None) → RollingFeatureCreator[source]

Fit method does nothing as no fitting is required.

Parameters:

X (pd.DataFrame) – Input DataFrame.
y (pd.Series, optional) – Target variable (ignored).

Returns:

Returns self.

Return type:

RollingFeatureCreator

transform(X: DataFrame) → DataFrame[source]

Creates rolling features.

Parameters:: X (pd.DataFrame) – Input DataFrame.
Returns:: DataFrame with new rolling feature columns.
Return type:: pd.DataFrame

class temporal_features.TimeDifferenceTransformer(column: str, new_column_name: str | None = None, periods: int = 1)[source]

Creates time difference between consecutive rows in a datetime column.

fit(X: DataFrame, y: Series | None = None) → TimeDifferenceTransformer[source]

Fit method does nothing as no fitting is required.

Parameters:

X (pd.DataFrame) – Input DataFrame.
y (pd.Series, optional) – Target variable (ignored).

Returns:

Returns self.

Return type:

TimeDifferenceTransformer

transform(X: DataFrame) → DataFrame[source]

Calculates time differences.

Parameters:: X (pd.DataFrame) – Input DataFrame.
Returns:: DataFrame with new time difference column.
Return type:: pd.DataFrame