File size: 17,636 Bytes
4aa0277 |
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 |
"""
Exploratory Data Analysis Agent - Handles comprehensive data analysis
"""
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
from scipy import stats
import warnings
warnings.filterwarnings('ignore')
class EDAAgent:
"""Agent for Exploratory Data Analysis"""
def __init__(self):
self.analysis_results = {}
def analyze_data(self, data, target_column=None):
"""
Comprehensive EDA analysis
Args:
data: Input DataFrame
target_column: Optional target variable for supervised analysis
Returns:
Dictionary containing comprehensive analysis results
"""
analysis = {}
# Basic statistics
analysis['basic_stats'] = self._basic_statistics(data)
# Correlation analysis
analysis['correlations'] = self._correlation_analysis(data)
# Distribution analysis
analysis['distributions'] = self._distribution_analysis(data)
# Feature insights
analysis['feature_insights'] = self._feature_insights(data)
# Target analysis (if target column provided)
if target_column and target_column in data.columns:
analysis['target_analysis'] = self._target_analysis(data, target_column)
# Data quality insights
analysis['data_quality'] = self._data_quality_insights(data)
return {
'status': 'success',
'analysis': analysis,
'visualization_recommendations': self._get_visualization_recommendations(data)
}
def _basic_statistics(self, data):
"""Generate comprehensive statistical summary"""
stats = {}
# Overall info
stats['shape'] = data.shape
stats['dtypes'] = data.dtypes.to_dict()
stats['memory_usage'] = f"{data.memory_usage(deep=True).sum() / 1024**2:.2f} MB"
# Numeric summary
numeric_data = data.select_dtypes(include=[np.number])
if not numeric_data.empty:
desc = numeric_data.describe()
stats['numeric_summary'] = desc.to_dict()
# Additional statistics
stats['numeric_extended'] = {}
for col in numeric_data.columns:
stats['numeric_extended'][col] = {
'variance': numeric_data[col].var(),
'skewness': numeric_data[col].skew(),
'kurtosis': numeric_data[col].kurtosis(),
'coefficient_of_variation': numeric_data[col].std() / numeric_data[col].mean() if numeric_data[col].mean() != 0 else np.inf
}
# Categorical summary
categorical_data = data.select_dtypes(include=['object', 'category'])
if not categorical_data.empty:
stats['categorical_summary'] = {}
for col in categorical_data.columns:
stats['categorical_summary'][col] = {
'unique_count': categorical_data[col].nunique(),
'most_frequent': categorical_data[col].mode().iloc[0] if len(categorical_data[col].mode()) > 0 else None,
'frequency_of_most_frequent': categorical_data[col].value_counts().iloc[0] if len(categorical_data[col]) > 0 else 0
}
# Missing values
stats['missing_values'] = data.isnull().sum().to_dict()
# Unique values count
stats['unique_values'] = {col: data[col].nunique() for col in data.columns}
return stats
def _correlation_analysis(self, data):
"""Analyze correlations between numeric variables"""
numeric_data = data.select_dtypes(include=[np.number])
if len(numeric_data.columns) < 2:
return {'message': 'Not enough numeric columns for correlation analysis'}
# Correlation matrix
corr_matrix = numeric_data.corr()
# Find strong correlations
strong_corr = []
for i in range(len(corr_matrix.columns)):
for j in range(i+1, len(corr_matrix.columns)):
corr_val = corr_matrix.iloc[i, j]
if not np.isnan(corr_val) and abs(corr_val) > 0.7:
strong_corr.append({
'var1': corr_matrix.columns[i],
'var2': corr_matrix.columns[j],
'correlation': corr_val,
'strength': 'very_strong' if abs(corr_val) > 0.9 else 'strong'
})
# Find moderate correlations
moderate_corr = []
for i in range(len(corr_matrix.columns)):
for j in range(i+1, len(corr_matrix.columns)):
corr_val = corr_matrix.iloc[i, j]
if not np.isnan(corr_val) and 0.3 <= abs(corr_val) <= 0.7:
moderate_corr.append({
'var1': corr_matrix.columns[i],
'var2': corr_matrix.columns[j],
'correlation': corr_val
})
return {
'correlation_matrix': corr_matrix.to_dict(),
'strong_correlations': strong_corr,
'moderate_correlations': moderate_corr[:10], # Limit to top 10
'summary': {
'total_pairs': len(corr_matrix.columns) * (len(corr_matrix.columns) - 1) // 2,
'strong_correlations_count': len(strong_corr),
'moderate_correlations_count': len(moderate_corr)
}
}
def _distribution_analysis(self, data):
"""Analyze distributions of all variables"""
distributions = {}
for col in data.columns:
col_info = {'column': col, 'dtype': str(data[col].dtype)}
if data[col].dtype in ['object', 'category']:
# Categorical distribution
value_counts = data[col].value_counts()
col_info.update({
'type': 'categorical',
'unique_count': len(value_counts),
'top_values': value_counts.head(10).to_dict(),
'entropy': stats.entropy(value_counts.values) if len(value_counts) > 1 else 0,
'most_frequent_percentage': (value_counts.iloc[0] / len(data)) * 100 if len(value_counts) > 0 else 0
})
else:
# Numerical distribution
col_data = data[col].dropna()
if len(col_data) > 0:
col_info.update({
'type': 'numerical',
'mean': col_data.mean(),
'median': col_data.median(),
'std': col_data.std(),
'min': col_data.min(),
'max': col_data.max(),
'skewness': col_data.skew(),
'kurtosis': col_data.kurtosis(),
'outliers_iqr': self._count_outliers_iqr(col_data),
'normality_test': self._test_normality(col_data)
})
distributions[col] = col_info
return distributions
def _feature_insights(self, data):
"""Generate feature insights and recommendations"""
insights = []
# Identify potential target variables
for col in data.columns:
unique_count = data[col].nunique()
if unique_count == 2:
insights.append({
'type': 'potential_target',
'feature': col,
'insight': f'{col} is binary - potential target for classification'
})
elif unique_count < 10 and data[col].dtype in ['object', 'string']:
insights.append({
'type': 'low_cardinality',
'feature': col,
'insight': f'{col} has low cardinality ({unique_count}) - good for classification target'
})
# Identify high cardinality categorical features
for col in data.select_dtypes(include=['object']).columns:
unique_count = data[col].nunique()
if unique_count > 50:
insights.append({
'type': 'high_cardinality',
'feature': col,
'insight': f'{col} has high cardinality ({unique_count}) - consider target encoding or grouping'
})
# Identify constant or near-constant features
for col in data.columns:
unique_count = data[col].nunique()
if unique_count == 1:
insights.append({
'type': 'constant_feature',
'feature': col,
'insight': f'{col} is constant - consider removing'
})
elif unique_count / len(data) < 0.01:
insights.append({
'type': 'near_constant',
'feature': col,
'insight': f'{col} is near-constant ({unique_count} unique values) - low information content'
})
# Identify features with many missing values
missing_threshold = 0.5
for col in data.columns:
missing_pct = data[col].isnull().sum() / len(data)
if missing_pct > missing_threshold:
insights.append({
'type': 'high_missing',
'feature': col,
'insight': f'{col} has {missing_pct:.1%} missing values - consider imputation or removal'
})
return insights
def _target_analysis(self, data, target_column):
"""Analyze target variable and its relationships"""
target = data[target_column]
analysis = {}
# Target distribution
if target.dtype in ['object', 'category']:
# Classification target
value_counts = target.value_counts()
analysis['type'] = 'classification'
analysis['classes'] = value_counts.to_dict()
analysis['class_balance'] = {
'balanced': max(value_counts) / min(value_counts) < 3,
'ratio': max(value_counts) / min(value_counts)
}
else:
# Regression target
analysis['type'] = 'regression'
analysis['distribution'] = {
'mean': target.mean(),
'median': target.median(),
'std': target.std(),
'skewness': target.skew(),
'kurtosis': target.kurtosis()
}
# Feature-target relationships
feature_relationships = []
other_features = [col for col in data.columns if col != target_column]
for feature in other_features[:20]: # Limit to first 20 features
if data[feature].dtype in [np.number]:
if analysis['type'] == 'classification':
# ANOVA F-test for numeric feature vs categorical target
try:
groups = [data[data[target_column] == cls][feature].dropna()
for cls in target.unique()]
f_stat, p_val = stats.f_oneway(*groups)
feature_relationships.append({
'feature': feature,
'test': 'ANOVA',
'f_statistic': f_stat,
'p_value': p_val,
'significant': p_val < 0.05
})
except:
pass
else:
# Correlation for numeric feature vs numeric target
corr, p_val = stats.pearsonr(data[feature].dropna(),
target[data[feature].notna()])
feature_relationships.append({
'feature': feature,
'test': 'Correlation',
'correlation': corr,
'p_value': p_val,
'significant': p_val < 0.05
})
analysis['feature_relationships'] = feature_relationships
return analysis
def _data_quality_insights(self, data):
"""Generate data quality insights"""
insights = []
# Overall data quality score
total_cells = data.shape[0] * data.shape[1]
missing_cells = data.isnull().sum().sum()
quality_score = (total_cells - missing_cells) / total_cells
insights.append({
'type': 'overall_quality',
'score': quality_score,
'interpretation': 'excellent' if quality_score > 0.95 else
'good' if quality_score > 0.85 else
'fair' if quality_score > 0.7 else 'poor'
})
# Duplicate rows
duplicate_count = data.duplicated().sum()
if duplicate_count > 0:
insights.append({
'type': 'duplicates',
'count': duplicate_count,
'percentage': (duplicate_count / len(data)) * 100
})
return insights
def _count_outliers_iqr(self, series):
"""Count outliers using IQR method"""
Q1 = series.quantile(0.25)
Q3 = series.quantile(0.75)
IQR = Q3 - Q1
lower_bound = Q1 - 1.5 * IQR
upper_bound = Q3 + 1.5 * IQR
outliers = series[(series < lower_bound) | (series > upper_bound)]
return len(outliers)
def _test_normality(self, series, max_samples=5000):
"""Test normality using Shapiro-Wilk test"""
try:
if len(series) > max_samples:
series_sample = series.sample(max_samples)
else:
series_sample = series
stat, p_value = stats.shapiro(series_sample)
return {
'test_statistic': stat,
'p_value': p_value,
'is_normal': p_value > 0.05
}
except:
return {'test_statistic': None, 'p_value': None, 'is_normal': None}
def _get_visualization_recommendations(self, data):
"""Generate visualization recommendations based on data characteristics"""
recommendations = []
numeric_cols = data.select_dtypes(include=[np.number]).columns
categorical_cols = data.select_dtypes(include=['object', 'category']).columns
# Distribution plots
if len(numeric_cols) > 0:
recommendations.append({
'type': 'histogram',
'purpose': 'Show distribution of numeric variables',
'columns': list(numeric_cols[:5])
})
recommendations.append({
'type': 'box_plot',
'purpose': 'Identify outliers in numeric variables',
'columns': list(numeric_cols[:5])
})
# Categorical plots
if len(categorical_cols) > 0:
recommendations.append({
'type': 'bar_chart',
'purpose': 'Show frequency of categorical variables',
'columns': list(categorical_cols[:5])
})
# Relationship plots
if len(numeric_cols) >= 2:
recommendations.append({
'type': 'correlation_heatmap',
'purpose': 'Show correlations between numeric variables',
'columns': list(numeric_cols)
})
recommendations.append({
'type': 'scatter_plot',
'purpose': 'Show relationships between numeric variables',
'columns': list(numeric_cols[:4])
})
# Mixed plots
if len(numeric_cols) > 0 and len(categorical_cols) > 0:
recommendations.append({
'type': 'grouped_box_plot',
'purpose': 'Show numeric distributions by categorical groups',
'numeric_columns': list(numeric_cols[:3]),
'categorical_columns': list(categorical_cols[:2])
})
return recommendations
def generate_insights_summary(self, analysis_results):
"""Generate a human-readable summary of key insights"""
if analysis_results['status'] != 'success':
return "Analysis failed"
analysis = analysis_results['analysis']
insights = []
# Basic stats insights
basic_stats = analysis['basic_stats']
insights.append(f"Dataset contains {basic_stats['shape'][0]:,} rows and {basic_stats['shape'][1]} columns")
# Missing values insight
missing_total = sum(basic_stats['missing_values'].values())
if missing_total > 0:
insights.append(f"Found {missing_total:,} missing values across the dataset")
# Correlation insights
if 'correlations' in analysis and 'strong_correlations' in analysis['correlations']:
strong_corr_count = len(analysis['correlations']['strong_correlations'])
if strong_corr_count > 0:
insights.append(f"Identified {strong_corr_count} strong correlations between variables")
# Feature insights
if 'feature_insights' in analysis:
feature_insights = analysis['feature_insights']
potential_targets = [i for i in feature_insights if i['type'] == 'potential_target']
if potential_targets:
insights.append(f"Found {len(potential_targets)} potential target variables for machine learning")
return insights
|