PCA with covariance
import numpy as npimport pandas as pdimport seaborn as snsimport matplotlib.pyplot as plt# Load datasetdataset = pd.read_csv('src/dataset.csv')pca = PCA(dataset, standardize=True, method='eig')normalized_dataset = pca.transformed_data# Covariance Matrix# bias =True, so dataset is normalized# rowvar = False, each column represents a variable, i.e., a feature. This way we compute the covariance of features as whole instead of the covariance of each rowcovariance_df = pd.DataFrame(data=np.cov(normalized_dataset, bias=True, rowvar=False), columns=dataset.columns)# Plot Covariance Matrixplt.subplots(figsize=(20, 20))sns.heatmap(covariance_df, cmap='Blues', linewidths=.7, annot=True, fmt='.2f', yticklabels=dataset.columns)plt.show()