arbitragelab.ml_approach.feature_expander

This module implements the Feature Expansion class.

Module Contents

Classes

FeatureExpander

Higher-order term Feature Expander implementation. The implementation consists

class FeatureExpander(methods: list = [], n_orders: int = 1)

Higher-order term Feature Expander implementation. The implementation consists of two major parts. The first part consists of using a collection of orthogonal polynomials’ coefficients, ordered from lowest order term to highest. The implemented series are [‘chebyshev’, ‘legendre’, ‘laguerre’, ‘power’] polynomials. The second part is a combinatorial version of feature crossing, which involves the generation of feature collections of the n order and multiplying them together. This can be used by adding [‘product’] in the ‘methods’ parameter in the constructor.

fit(frame: pandas.DataFrame)

Stores the dataset inside the class object.

Parameters:

frame – (pd.DataFrame) Dataset to store.

transform() pandas.DataFrame

Returns the original dataframe with features requested from the ‘methods’ parameter in the constructor.

Returns:

(pd.DataFrame) Original DataFrame with the expanded values appended to it.