如何在Python或R中获得选择主成分个数的BIC/AIC图

import numpy as np
from sklearn.decomposition import PCA
import matplotlib.pyplot as plt

# 准备数据
data = ...

# 进行主成分分析
pca = PCA()
pca.fit(data)

# 计算BIC和AIC值
n_components = range(1, len(data.columns)+1)
bic = []
aic = []
for n in n_components:
    pca.n_components = n
    pca.fit(data)
    log_likelihood = -np.sum(np.log(pca.explained_variance_))
    num_params = n * (n + 1) / 2
    bic.append(log_likelihood + 0.5 * num_params * np.log(len(data)))
    aic.append(log_likelihood + num_params)

# 绘制BIC/AIC图
plt.plot(n_components, bic, label='BIC')
plt.plot(n_components, aic, label='AIC')
plt.xlabel('Number of Components')
plt.ylabel('BIC/AIC Value')
plt.title('BIC/AIC Plot for Selecting Number of Components')
plt.legend()
plt.show()

library(stats)
library(factoextra)

# 准备数据
data <- ...

# 进行主成分分析
pca <- princomp(data)

# 计算BIC和AIC值
n_components <- 1:length(data)
bic <- numeric(length(n_components))
aic <- numeric(length(n_components))
for (i in n_components) {
  pca_i <- princomp(data, cor = TRUE, scores = TRUE)
  log_likelihood <- -sum(log(pca_i$sdev^2))
  num_params <- i * (i + 1) / 2
  bic[i] <- log_likelihood + 0.5 * num_params * log(length(data))
  aic[i] <- log_likelihood + num_params
}

# 绘制BIC/AIC图
fviz_screeplot(pca, addlabels = TRUE, ylim = c(0, 1), select = "bic")
fviz_screeplot(pca, addlabels = TRUE, ylim = c(0, 1), select = "aic")