Laplacian Score

An implementation of laplacian score by Python since all the code on GitHub are either too complicated or unavailable

Reference He, Xiaofei, Deng Cai, and Partha Niyogi. "Laplacian score for feature selection." Advances in neural information processing systems  18 (2005).

Usage

laplacian_score(
    df_arr: numpy array,
    label: None,
    **kwargs
)

Arguments:

• df_arr: A numpy array represent data
• label: Default=None, Label if the data is a supervised data.
• k_nearest: Default=5, if the data is an unsupervised data, use k_nearest to find the edge of distance graph, this parameter takes no action if label parameter exists.

Example

from sklearn import datasets
import pandas as pd
import numpy as np
iris = datasets.load_iris()
df = pd.DataFrame(iris["data"], columns=iris["feature_names"])
target = iris["target"]
print(df.head())

sepal length (cm)	sepal width (cm)	petal length (cm)	petal width (cm)
5.1	3.5	1.4	0.2
4.9	3.0	1.4	0.2
4.7	3.2	1.3	0.2

No lable data (Unsupervised)

>>> laplacian_score(np.array(df), k_nearest=3)
array([0.04132189, 0.17426192, 0.01415396, 0.03982228])

The value above corresponding to laplacian score of each features, the smaller the score, the higher chance be selected

Labled data (Supervised)

>>> laplacian_score(np.array(df), label=target)
array([0.60421927, 0.78376157, 0.1138573 , 0.10572175])