File size: 18,292 Bytes
26a8ea5 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 |
"""
Economic Segmentation Module
Advanced clustering analysis for economic time series and time periods
"""
import logging
from typing import Dict, List, Optional, Tuple, Union
import numpy as np
import pandas as pd
from sklearn.cluster import KMeans, AgglomerativeClustering
from sklearn.decomposition import PCA
from sklearn.manifold import TSNE
from sklearn.metrics import silhouette_score, calinski_harabasz_score
from sklearn.preprocessing import StandardScaler
from scipy.cluster.hierarchy import dendrogram, linkage, fcluster
from scipy.spatial.distance import pdist, squareform
logger = logging.getLogger(__name__)
class EconomicSegmentation:
"""
Advanced economic segmentation using clustering techniques
for both time periods and economic series
"""
def __init__(self, data: pd.DataFrame):
"""
Initialize segmentation with economic data
Args:
data: DataFrame with economic indicators
"""
self.data = data.copy()
self.scaler = StandardScaler()
self.clusters = {}
self.cluster_analysis = {}
def prepare_time_period_data(self, indicators: List[str] = None,
window_size: int = 4) -> pd.DataFrame:
"""
Prepare time period data for clustering
Args:
indicators: List of indicators to use. If None, use all numeric columns
window_size: Rolling window size for feature extraction
Returns:
DataFrame with time period features
"""
if indicators is None:
indicators = self.data.select_dtypes(include=[np.number]).columns.tolist()
# Calculate growth rates for economic indicators
growth_data = self.data[indicators].pct_change().dropna()
# Extract features for each time period
features = []
feature_names = []
for indicator in indicators:
# Rolling statistics
features.extend([
growth_data[indicator].rolling(window_size).mean(),
growth_data[indicator].rolling(window_size).std(),
growth_data[indicator].rolling(window_size).min(),
growth_data[indicator].rolling(window_size).max(),
growth_data[indicator].rolling(window_size).skew(),
growth_data[indicator].rolling(window_size).kurt()
])
feature_names.extend([
f"{indicator}_mean", f"{indicator}_std", f"{indicator}_min",
f"{indicator}_max", f"{indicator}_skew", f"{indicator}_kurt"
])
# Create feature matrix
feature_df = pd.concat(features, axis=1)
feature_df.columns = feature_names
feature_df = feature_df.dropna()
return feature_df
def prepare_series_data(self, indicators: List[str] = None) -> pd.DataFrame:
"""
Prepare series data for clustering (clustering the indicators themselves)
Args:
indicators: List of indicators to use. If None, use all numeric columns
Returns:
DataFrame with series features
"""
if indicators is None:
indicators = self.data.select_dtypes(include=[np.number]).columns.tolist()
# Calculate growth rates
growth_data = self.data[indicators].pct_change().dropna()
# Extract features for each series
series_features = {}
for indicator in indicators:
series = growth_data[indicator].dropna()
# Statistical features
series_features[indicator] = {
'mean': series.mean(),
'std': series.std(),
'min': series.min(),
'max': series.max(),
'skew': series.skew(),
'kurt': series.kurtosis(),
'autocorr_1': series.autocorr(lag=1),
'autocorr_4': series.autocorr(lag=4),
'volatility': series.rolling(12).std().mean(),
'trend': np.polyfit(range(len(series)), series, 1)[0]
}
return pd.DataFrame(series_features).T
def find_optimal_clusters(self, data: pd.DataFrame, max_clusters: int = 10,
method: str = 'kmeans') -> Dict:
"""
Find optimal number of clusters using elbow method and silhouette analysis
Args:
data: Feature data for clustering
max_clusters: Maximum number of clusters to test
method: Clustering method ('kmeans' or 'hierarchical')
Returns:
Dictionary with optimal cluster analysis
"""
if len(data) < max_clusters:
max_clusters = len(data) - 1
inertias = []
silhouette_scores = []
calinski_scores = []
for k in range(2, max_clusters + 1):
try:
if method == 'kmeans':
kmeans = KMeans(n_clusters=k, random_state=42, n_init=10)
labels = kmeans.fit_predict(data)
inertias.append(kmeans.inertia_)
else:
clustering = AgglomerativeClustering(n_clusters=k)
labels = clustering.fit_predict(data)
inertias.append(0) # Not applicable for hierarchical
# Calculate scores
if len(np.unique(labels)) > 1:
silhouette_scores.append(silhouette_score(data, labels))
calinski_scores.append(calinski_harabasz_score(data, labels))
else:
silhouette_scores.append(0)
calinski_scores.append(0)
except Exception as e:
logger.warning(f"Failed to cluster with k={k}: {e}")
inertias.append(0)
silhouette_scores.append(0)
calinski_scores.append(0)
# Find optimal k using silhouette score
optimal_k_silhouette = np.argmax(silhouette_scores) + 2
optimal_k_calinski = np.argmax(calinski_scores) + 2
# Elbow method (for k-means)
if method == 'kmeans' and len(inertias) > 1:
# Calculate second derivative to find elbow
second_derivative = np.diff(np.diff(inertias))
optimal_k_elbow = np.argmin(second_derivative) + 3
else:
optimal_k_elbow = optimal_k_silhouette
return {
'inertias': inertias,
'silhouette_scores': silhouette_scores,
'calinski_scores': calinski_scores,
'optimal_k_silhouette': optimal_k_silhouette,
'optimal_k_calinski': optimal_k_calinski,
'optimal_k_elbow': optimal_k_elbow,
'recommended_k': optimal_k_silhouette # Use silhouette as primary
}
def cluster_time_periods(self, indicators: List[str] = None,
n_clusters: int = None, method: str = 'kmeans',
window_size: int = 4) -> Dict:
"""
Cluster time periods based on economic activity patterns
Args:
indicators: List of indicators to use
n_clusters: Number of clusters. If None, auto-detect
method: Clustering method ('kmeans' or 'hierarchical')
window_size: Rolling window size for feature extraction
Returns:
Dictionary with clustering results
"""
# Prepare data
feature_df = self.prepare_time_period_data(indicators, window_size)
# Scale features
scaled_data = self.scaler.fit_transform(feature_df)
scaled_df = pd.DataFrame(scaled_data, index=feature_df.index, columns=feature_df.columns)
# Find optimal clusters if not specified
if n_clusters is None:
cluster_analysis = self.find_optimal_clusters(scaled_df, method=method)
n_clusters = cluster_analysis['recommended_k']
logger.info(f"Auto-detected optimal clusters: {n_clusters}")
# Perform clustering
if method == 'kmeans':
clustering = KMeans(n_clusters=n_clusters, random_state=42, n_init=10)
else:
clustering = AgglomerativeClustering(n_clusters=n_clusters)
cluster_labels = clustering.fit_predict(scaled_df)
# Add cluster labels to original data
result_df = feature_df.copy()
result_df['cluster'] = cluster_labels
# Analyze clusters
cluster_analysis = self.analyze_clusters(result_df, 'cluster')
# Dimensionality reduction for visualization
pca = PCA(n_components=2)
pca_data = pca.fit_transform(scaled_data)
tsne = TSNE(n_components=2, random_state=42, perplexity=min(30, len(scaled_data)-1))
tsne_data = tsne.fit_transform(scaled_data)
return {
'data': result_df,
'cluster_labels': cluster_labels,
'cluster_analysis': cluster_analysis,
'pca_data': pca_data,
'tsne_data': tsne_data,
'feature_importance': dict(zip(feature_df.columns, np.abs(pca.components_[0]))),
'n_clusters': n_clusters,
'method': method
}
def cluster_economic_series(self, indicators: List[str] = None,
n_clusters: int = None, method: str = 'kmeans') -> Dict:
"""
Cluster economic series based on their characteristics
Args:
indicators: List of indicators to use
n_clusters: Number of clusters. If None, auto-detect
method: Clustering method ('kmeans' or 'hierarchical')
Returns:
Dictionary with clustering results
"""
# Prepare data
series_df = self.prepare_series_data(indicators)
# Scale features
scaled_data = self.scaler.fit_transform(series_df)
scaled_df = pd.DataFrame(scaled_data, index=series_df.index, columns=series_df.columns)
# Find optimal clusters if not specified
if n_clusters is None:
cluster_analysis = self.find_optimal_clusters(scaled_df, method=method)
n_clusters = cluster_analysis['recommended_k']
logger.info(f"Auto-detected optimal clusters: {n_clusters}")
# Perform clustering
if method == 'kmeans':
clustering = KMeans(n_clusters=n_clusters, random_state=42, n_init=10)
else:
clustering = AgglomerativeClustering(n_clusters=n_clusters)
cluster_labels = clustering.fit_predict(scaled_df)
# Add cluster labels
result_df = series_df.copy()
result_df['cluster'] = cluster_labels
# Analyze clusters
cluster_analysis = self.analyze_clusters(result_df, 'cluster')
# Dimensionality reduction for visualization
pca = PCA(n_components=2)
pca_data = pca.fit_transform(scaled_data)
tsne = TSNE(n_components=2, random_state=42, perplexity=min(30, len(scaled_data)-1))
tsne_data = tsne.fit_transform(scaled_data)
return {
'data': result_df,
'cluster_labels': cluster_labels,
'cluster_analysis': cluster_analysis,
'pca_data': pca_data,
'tsne_data': tsne_data,
'feature_importance': dict(zip(series_df.columns, np.abs(pca.components_[0]))),
'n_clusters': n_clusters,
'method': method
}
def analyze_clusters(self, data: pd.DataFrame, cluster_col: str) -> Dict:
"""
Analyze cluster characteristics
Args:
data: DataFrame with cluster labels
cluster_col: Name of cluster column
Returns:
Dictionary with cluster analysis
"""
feature_cols = [col for col in data.columns if col != cluster_col]
cluster_analysis = {}
for cluster_id in data[cluster_col].unique():
cluster_data = data[data[cluster_col] == cluster_id]
cluster_analysis[cluster_id] = {
'size': len(cluster_data),
'percentage': len(cluster_data) / len(data) * 100,
'features': {}
}
# Analyze each feature
for feature in feature_cols:
feature_data = cluster_data[feature]
cluster_analysis[cluster_id]['features'][feature] = {
'mean': feature_data.mean(),
'std': feature_data.std(),
'min': feature_data.min(),
'max': feature_data.max(),
'median': feature_data.median()
}
return cluster_analysis
def perform_hierarchical_clustering(self, data: pd.DataFrame,
method: str = 'ward',
distance_threshold: float = None) -> Dict:
"""
Perform hierarchical clustering with dendrogram analysis
Args:
data: Feature data for clustering
method: Linkage method ('ward', 'complete', 'average', 'single')
distance_threshold: Distance threshold for cutting dendrogram
Returns:
Dictionary with hierarchical clustering results
"""
# Scale data
scaled_data = self.scaler.fit_transform(data)
# Calculate linkage matrix
if method == 'ward':
linkage_matrix = linkage(scaled_data, method=method)
else:
# For non-ward methods, we need to provide distance matrix
distance_matrix = pdist(scaled_data)
linkage_matrix = linkage(distance_matrix, method=method)
# Determine number of clusters
if distance_threshold is None:
# Use elbow method on distance
distances = linkage_matrix[:, 2]
second_derivative = np.diff(np.diff(distances))
optimal_threshold = distances[np.argmax(second_derivative) + 1]
else:
optimal_threshold = distance_threshold
# Get cluster labels
cluster_labels = fcluster(linkage_matrix, optimal_threshold, criterion='distance')
# Analyze clusters
result_df = data.copy()
result_df['cluster'] = cluster_labels
cluster_analysis = self.analyze_clusters(result_df, 'cluster')
return {
'linkage_matrix': linkage_matrix,
'cluster_labels': cluster_labels,
'distance_threshold': optimal_threshold,
'cluster_analysis': cluster_analysis,
'data': result_df,
'method': method
}
def generate_segmentation_report(self, time_period_clusters: Dict = None,
series_clusters: Dict = None) -> str:
"""
Generate comprehensive segmentation report
Args:
time_period_clusters: Results from time period clustering
series_clusters: Results from series clustering
Returns:
Formatted report string
"""
report = "ECONOMIC SEGMENTATION REPORT\n"
report += "=" * 50 + "\n\n"
if time_period_clusters:
report += "TIME PERIOD CLUSTERING\n"
report += "-" * 30 + "\n"
report += f"Method: {time_period_clusters['method']}\n"
report += f"Number of Clusters: {time_period_clusters['n_clusters']}\n"
report += f"Total Periods: {len(time_period_clusters['data'])}\n\n"
# Cluster summary
cluster_analysis = time_period_clusters['cluster_analysis']
for cluster_id, analysis in cluster_analysis.items():
report += f"Cluster {cluster_id}:\n"
report += f" Size: {analysis['size']} periods ({analysis['percentage']:.1f}%)\n"
# Top features for this cluster
if 'feature_importance' in time_period_clusters:
features = time_period_clusters['feature_importance']
top_features = sorted(features.items(), key=lambda x: x[1], reverse=True)[:5]
report += f" Top Features: {', '.join([f[0] for f in top_features])}\n"
report += "\n"
if series_clusters:
report += "ECONOMIC SERIES CLUSTERING\n"
report += "-" * 30 + "\n"
report += f"Method: {series_clusters['method']}\n"
report += f"Number of Clusters: {series_clusters['n_clusters']}\n"
report += f"Total Series: {len(series_clusters['data'])}\n\n"
# Cluster summary
cluster_analysis = series_clusters['cluster_analysis']
for cluster_id, analysis in cluster_analysis.items():
report += f"Cluster {cluster_id}:\n"
report += f" Size: {analysis['size']} series ({analysis['percentage']:.1f}%)\n"
# Series in this cluster
cluster_series = series_clusters['data'][series_clusters['data']['cluster'] == cluster_id]
series_names = cluster_series.index.tolist()
report += f" Series: {', '.join(series_names)}\n"
# Top features for this cluster
if 'feature_importance' in series_clusters:
features = series_clusters['feature_importance']
top_features = sorted(features.items(), key=lambda x: x[1], reverse=True)[:5]
report += f" Top Features: {', '.join([f[0] for f in top_features])}\n"
report += "\n"
return report |