302 lines
12 KiB
Python
302 lines
12 KiB
Python
import gc
|
||
import os
|
||
|
||
import numpy as np
|
||
import pandas as pd
|
||
import matplotlib.pyplot as plt
|
||
import seaborn as sns
|
||
from sklearn.decomposition import PCA
|
||
from sklearn.cluster import KMeans
|
||
|
||
|
||
def generate_correlation_heatmap(self):
|
||
"""生成相关性热力图"""
|
||
fig = None
|
||
try:
|
||
self._log_step("Generating correlation heatmap...")
|
||
|
||
if not hasattr(self, 'data') or self.data is None or len(self.data.columns) <= 1:
|
||
self._log_step("Not enough data for correlation analysis", "warning")
|
||
return None, "Not enough data", None
|
||
|
||
# 计算相关性矩阵
|
||
numeric_cols = self.data.select_dtypes(include=[np.number]).columns
|
||
corr_matrix = self.data[numeric_cols].corr()
|
||
summary = "Correlation matrix calculated"
|
||
|
||
if not self.generate_plots:
|
||
self._log_step("Correlation analysis completed (data only)", "success")
|
||
# 替换NaN为None以兼容JSON
|
||
return None, summary, corr_matrix.where(pd.notnull(corr_matrix), None)
|
||
|
||
# 创建热力图
|
||
fig = plt.figure(figsize=(8, 6), dpi=100)
|
||
sns.heatmap(
|
||
corr_matrix,
|
||
annot=True,
|
||
fmt=".2f",
|
||
cmap='coolwarm',
|
||
center=0,
|
||
square=True,
|
||
cbar_kws={"shrink": 0.8},
|
||
)
|
||
plt.title('Correlation Heatmap')
|
||
plt.tight_layout()
|
||
|
||
# 保存图片
|
||
img_path = os.path.join(self.temp_dir.name, 'correlation_heatmap.png')
|
||
try:
|
||
plt.savefig(img_path, dpi=100, bbox_inches='tight', format='png')
|
||
except Exception as save_err:
|
||
self._log_step(f"Save error: {save_err}", "error")
|
||
return None, f"Save error: {str(save_err)[:100]}", corr_matrix.where(pd.notnull(corr_matrix), None)
|
||
finally:
|
||
plt.close(fig)
|
||
gc.collect()
|
||
|
||
self._log_step("Correlation heatmap generated", "success")
|
||
|
||
return img_path, summary, corr_matrix.where(pd.notnull(corr_matrix), None)
|
||
|
||
except Exception as e:
|
||
self._log_step(f"Correlation heatmap failed: {str(e)[:100]}", "error")
|
||
if fig is not None:
|
||
try:
|
||
plt.close(fig)
|
||
except Exception:
|
||
pass
|
||
return None, f"Correlation heatmap failed: {str(e)[:100]}", None
|
||
|
||
|
||
def generate_pca_scree_plot(self):
|
||
"""生成PCA碎石图"""
|
||
try:
|
||
self._log_step("Generating PCA scree plot...")
|
||
|
||
if hasattr(self, 'scaled_data') and self.scaled_data is not None:
|
||
pca = PCA()
|
||
pca.fit(self.scaled_data)
|
||
|
||
explained_variance = pca.explained_variance_ratio_
|
||
cumulative_variance = np.cumsum(explained_variance)
|
||
|
||
# 准备数据
|
||
scree_data = pd.DataFrame({
|
||
'component': range(1, len(explained_variance) + 1),
|
||
'explained_variance': explained_variance,
|
||
'cumulative_variance': cumulative_variance,
|
||
})
|
||
|
||
summary = (
|
||
"PCA碎石图生成完成，前2个主成分解释 "
|
||
f"{cumulative_variance[min(1, len(cumulative_variance) - 1)]:.2%} 方差"
|
||
)
|
||
|
||
if not self.generate_plots:
|
||
self._log_step("PCA scree data generated", "success")
|
||
return None, summary, scree_data
|
||
|
||
# 创建碎石图
|
||
plt.figure(figsize=(10, 6))
|
||
|
||
# 绘制碎石图
|
||
plt.subplot(1, 2, 1)
|
||
plt.plot(range(1, len(explained_variance) + 1), explained_variance, 'bo-')
|
||
plt.title('PCA碎石图')
|
||
plt.xlabel('主成分')
|
||
plt.ylabel('解释方差比例')
|
||
plt.grid(True, alpha=0.3)
|
||
|
||
# 绘制累积方差图
|
||
plt.subplot(1, 2, 2)
|
||
plt.plot(range(1, len(cumulative_variance) + 1), cumulative_variance, 'ro-')
|
||
plt.title('累积解释方差')
|
||
plt.xlabel('主成分数量')
|
||
plt.ylabel('累积方差比例')
|
||
plt.axhline(y=0.85, color='g', linestyle='--', label='85% 方差')
|
||
plt.legend()
|
||
plt.grid(True, alpha=0.3)
|
||
|
||
plt.tight_layout()
|
||
img_path = os.path.join(self.temp_dir.name, 'pca_scree_plot.png')
|
||
plt.savefig(img_path, dpi=150, bbox_inches='tight')
|
||
plt.close()
|
||
|
||
self._log_step("PCA scree plot generated", "success")
|
||
|
||
return img_path, summary, scree_data
|
||
|
||
self._log_step("No scaled data available for PCA scree plot", "warning")
|
||
return None, "没有标准化数据可用于PCA碎石图", None
|
||
|
||
except Exception as e:
|
||
self._log_step(f"PCA scree plot failed: {e}", "error")
|
||
return None, f"PCA碎石图生成失败: {e}", None
|
||
|
||
|
||
def perform_pca_analysis(self):
|
||
"""执行主成分分析"""
|
||
try:
|
||
self._log_step("Performing PCA analysis...")
|
||
|
||
if hasattr(self, 'scaled_data') and self.scaled_data is not None and len(self.scaled_data.columns) > 1:
|
||
pca = PCA(n_components=2)
|
||
principal_components = pca.fit_transform(self.scaled_data)
|
||
|
||
summary = (
|
||
"PCA analysis completed, explained variance: "
|
||
f"{pca.explained_variance_ratio_[0]:.2%} + {pca.explained_variance_ratio_[1]:.2%}"
|
||
)
|
||
|
||
if not self.generate_plots:
|
||
self._log_step("PCA analysis completed (data only)", "success")
|
||
pca_df = pd.DataFrame(data=principal_components, columns=['PC1', 'PC2'])
|
||
pca_df['timestamp'] = self.data.index.astype(str)
|
||
return None, summary, pca_df
|
||
|
||
pca_df = pd.DataFrame(data=principal_components, columns=['PC1', 'PC2'])
|
||
pca_df['timestamp'] = self.data.index.astype(str)
|
||
|
||
# 创建PCA散点图
|
||
plt.figure(figsize=(8, 6))
|
||
plt.scatter(principal_components[:, 0], principal_components[:, 1], alpha=0.7)
|
||
plt.xlabel(f'PC1 ({pca.explained_variance_ratio_[0]:.2%})')
|
||
plt.ylabel(f'PC2 ({pca.explained_variance_ratio_[1]:.2%})')
|
||
plt.title('Principal Component Analysis (PCA)')
|
||
plt.grid(True, alpha=0.3)
|
||
plt.tight_layout()
|
||
|
||
# 保存图片
|
||
img_path = os.path.join(self.temp_dir.name, 'pca_analysis.png')
|
||
plt.savefig(img_path, dpi=150, bbox_inches='tight')
|
||
plt.close()
|
||
|
||
self._log_step("PCA analysis completed", "success")
|
||
|
||
return img_path, summary, pca_df
|
||
|
||
self._log_step("Not enough data for PCA analysis", "warning")
|
||
return None, "Not enough data for PCA analysis", None
|
||
|
||
except Exception as e:
|
||
self._log_step(f"PCA analysis failed: {e}", "error")
|
||
return None, f"PCA analysis failed: {e}", None
|
||
|
||
|
||
def perform_clustering_analysis(self):
|
||
"""执行聚类分析"""
|
||
try:
|
||
self._log_step("Performing clustering analysis...")
|
||
|
||
if hasattr(self, 'scaled_data') and self.scaled_data is not None and len(self.scaled_data.columns) > 1:
|
||
kmeans = KMeans(n_clusters=3, random_state=42)
|
||
clusters = kmeans.fit_predict(self.scaled_data)
|
||
|
||
summary = f"Clustering analysis completed, found {len(np.unique(clusters))} clusters"
|
||
|
||
if not self.generate_plots:
|
||
self._log_step("Clustering analysis completed (data only)", "success")
|
||
cluster_df = pd.DataFrame({'cluster': clusters})
|
||
cluster_df['timestamp'] = self.data.index.astype(str)
|
||
return None, summary, cluster_df
|
||
|
||
cluster_df = pd.DataFrame({'cluster': clusters})
|
||
cluster_df['timestamp'] = self.data.index.astype(str)
|
||
|
||
# 如果数据是2D的，直接绘制聚类结果
|
||
if len(self.scaled_data.columns) >= 2:
|
||
plt.figure(figsize=(8, 6))
|
||
plt.scatter(
|
||
self.scaled_data.iloc[:, 0],
|
||
self.scaled_data.iloc[:, 1],
|
||
c=clusters,
|
||
cmap='viridis',
|
||
alpha=0.7,
|
||
)
|
||
plt.xlabel(self.scaled_data.columns[0])
|
||
plt.ylabel(self.scaled_data.columns[1])
|
||
plt.title('Clustering Analysis')
|
||
plt.colorbar(label='Cluster')
|
||
plt.tight_layout()
|
||
else:
|
||
# 对于高维数据，使用PCA降维后可视化
|
||
pca = PCA(n_components=2)
|
||
reduced_data = pca.fit_transform(self.scaled_data)
|
||
|
||
plt.figure(figsize=(8, 6))
|
||
plt.scatter(reduced_data[:, 0], reduced_data[:, 1], c=clusters, cmap='viridis', alpha=0.7)
|
||
plt.xlabel('PC1')
|
||
plt.ylabel('PC2')
|
||
plt.title('Clustering Analysis (PCA Reduced)')
|
||
plt.colorbar(label='Cluster')
|
||
plt.tight_layout()
|
||
|
||
# 保存图片
|
||
img_path = os.path.join(self.temp_dir.name, 'clustering_analysis.png')
|
||
plt.savefig(img_path, dpi=150, bbox_inches='tight')
|
||
plt.close()
|
||
|
||
self._log_step("Clustering analysis completed", "success")
|
||
|
||
return img_path, summary, cluster_df
|
||
|
||
self._log_step("Not enough data for clustering analysis", "warning")
|
||
return None, "Not enough data for clustering analysis", None
|
||
|
||
except Exception as e:
|
||
self._log_step(f"Clustering analysis failed: {e}", "error")
|
||
return None, f"Clustering analysis failed: {e}", None
|
||
|
||
|
||
def perform_factor_analysis(self):
|
||
"""执行因子分析"""
|
||
try:
|
||
self._log_step("Performing factor analysis...")
|
||
|
||
if hasattr(self, 'scaled_data') and self.scaled_data is not None and len(self.scaled_data.columns) > 1:
|
||
from sklearn.decomposition import FactorAnalysis
|
||
|
||
fa = FactorAnalysis(n_components=2, random_state=42)
|
||
factors = fa.fit_transform(self.scaled_data)
|
||
|
||
summary = "因子分析完成，提取了2个主要因子"
|
||
|
||
if not self.generate_plots:
|
||
self._log_step("Factor analysis completed (data only)", "success")
|
||
factor_df = pd.DataFrame(data=factors, columns=['Factor1', 'Factor2'])
|
||
factor_df['timestamp'] = self.data.index.astype(str)
|
||
return None, summary, factor_df
|
||
|
||
factor_df = pd.DataFrame(data=factors, columns=['Factor1', 'Factor2'])
|
||
factor_df['timestamp'] = self.data.index.astype(str)
|
||
|
||
# 创建因子分析图
|
||
plt.figure(figsize=(10, 8))
|
||
plt.scatter(factors[:, 0], factors[:, 1], alpha=0.7)
|
||
plt.xlabel('Factor 1')
|
||
plt.ylabel('Factor 2')
|
||
plt.title('Factor Analysis')
|
||
plt.grid(True, alpha=0.3)
|
||
|
||
# 添加因子载荷
|
||
for i, (x, y) in enumerate(factors[:10]): # 只显示前10个点
|
||
plt.annotate(str(i), (x, y), xytext=(5, 5), textcoords='offset points', fontsize=8)
|
||
|
||
plt.tight_layout()
|
||
|
||
# 保存图片
|
||
img_path = os.path.join(self.temp_dir.name, 'factor_analysis.png')
|
||
plt.savefig(img_path, dpi=150, bbox_inches='tight')
|
||
plt.close()
|
||
|
||
self._log_step("Factor analysis completed", "success")
|
||
|
||
return img_path, summary, factor_df
|
||
|
||
self._log_step("Not enough data for factor analysis", "warning")
|
||
return None, "数据不足，无法进行因子分析", None
|
||
|
||
except Exception as e:
|
||
self._log_step(f"Factor analysis failed: {e}", "error")
|
||
return None, f"因子分析失败: {e}", None
|