主成分分析

gistfile1.txt

import numpy as np
from sklearn.preprocessing import StandardScaler
from sklearn.decomposition import PCA

# Excluding the target variable 'Diabetes_binary' for PCA
features = data.drop('Diabetes_binary', axis=1)

# Scaling the data
scaler = StandardScaler()
scaled_features = scaler.fit_transform(features)

# Performing PCA
pca = PCA()
principal_components = pca.fit_transform(scaled_features)

# Explained variance ratio for each principal component
explained_variance_ratio = pca.explained_variance_ratio_

# Calculating the cumulative variance
cumulative_variance = explained_variance_ratio.cumsum()

# Finding the number of components to reach 80% cumulative variance
n_components_80 = (cumulative_variance >= 0.80).argmax() + 1

# Performing PCA with the desired number of components
pca_80 = PCA(n_components=n_components_80)
principal_components_80 = pca_80.fit_transform(scaled_features)

# Creating a DataFrame for the principal components
principal_df_80 = pd.DataFrame(data=principal_components_80, columns=[f'PC{i}' for i in range(n_components_80)])

# Calculating loadings (correlation between original features and the selected principal components)
loadings_80 = pca_80.components_.T * np.sqrt(pca_80.explained_variance_)

# Creating a DataFrame for the loadings
loadings_df_80 = pd.DataFrame(data=loadings_80, columns=[f'PC{i}' for i in range(n_components_80)], index=features.columns)

# Transposing the loadings DataFrame to swap rows and columns
loadings_df_80_transposed = loadings_df_80.transpose()

# Display the transposed loadings DataFrame
loadings_df_80_transposed