File size: 10,228 Bytes
9deebf2 | 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 | #!/usr/bin/env python3
"""Audit: Users (Survey Responses) dataset."""
import sys
sys.path.insert(0, str(__import__('pathlib').Path(__file__).resolve().parent))
from _common import *
def main():
output_dir = setup_output_dir('users')
print_section('USERS DATASET AUDIT')
print(f'Audit started: {datetime.now()}')
df = load_parquet('users')
# --- Overview ---
print_section('DATASET OVERVIEW')
df.info()
# --- Missing data ---
print_section('MISSING DATA SUMMARY')
missing = df.isnull().sum()
missing_pct = (missing / len(df) * 100).round(2)
missing_df = pd.DataFrame({
'Missing Count': missing,
'Missing %': missing_pct
}).sort_values('Missing %', ascending=False)
print(missing_df[missing_df['Missing Count'] > 0])
# --- Psi/Hemi summaries ---
psi_cols = [f'psi_{i:02d}' for i in range(1, 16)]
hemi_cols = [f'hemi_{i:02d}' for i in range(1, 11)]
print_section('SURVEY QUESTION SUMMARIES')
print('PSI QUESTIONS SUMMARY (1-5 scale)')
print(df[psi_cols].describe())
print('\nHEMI QUESTIONS SUMMARY (1-5 scale)')
print(df[hemi_cols].describe())
# --- Temporal ---
df['timestamp'] = pd.to_datetime(df['timestamp'])
df['date'] = df['timestamp'].dt.date
df['year'] = df['timestamp'].dt.year
df['month'] = df['timestamp'].dt.to_period('M')
print_section('TEMPORAL ANALYSIS')
print(f'First record: {df["timestamp"].min()}')
print(f'Last record: {df["timestamp"].max()}')
print(f'Time span: {(df["timestamp"].max() - df["timestamp"].min()).days} days')
print(f'Years: {df["year"].min()} to {df["year"].max()}')
yearly_counts = df['year'].value_counts().sort_index()
print('\nYEARLY DISTRIBUTION')
print(yearly_counts)
plt.figure(figsize=(14, 6))
yearly_counts.plot(kind='bar', color='steelblue')
plt.title('Users Dataset Audit - Yearly Distribution', fontsize=14, fontweight='bold')
plt.xlabel('Year')
plt.ylabel('Number of Users')
plt.xticks(rotation=45)
plt.grid(axis='y', alpha=0.3)
plt.tight_layout()
save_fig(output_dir, 'yearly_distribution')
# Monthly continuity
monthly_counts = df.groupby('month').size()
plt.figure(figsize=(16, 6))
monthly_counts.plot(kind='line', marker='o', markersize=3)
plt.title('Survey Responses by Month (Temporal Continuity Check)', fontsize=14, fontweight='bold')
plt.xlabel('Month')
plt.ylabel('Number of Users')
plt.grid(alpha=0.3)
plt.tight_layout()
save_fig(output_dir, 'monthly_continuity')
all_months = pd.period_range(start=monthly_counts.index.min(),
end=monthly_counts.index.max(), freq='M')
missing_months = all_months.difference(monthly_counts.index)
if len(missing_months) > 0:
print(f'\n{len(missing_months)} months with ZERO responses:')
for month in missing_months[:10]:
print(f' - {month}')
if len(missing_months) > 10:
print(f' ... and {len(missing_months) - 10} more')
else:
print('\nNo missing months - data is temporally continuous')
# --- Deduplication check ---
print_section('DEDUPLICATION CHECK')
uid_col = 'username_hash' if 'username_hash' in df.columns else 'username'
print(f'Total rows: {len(df):,}')
print(f'Unique {uid_col}: {df[uid_col].nunique():,}')
if 'username' in df.columns:
duplicates = df[df.duplicated(subset=['username'], keep=False)]
if len(duplicates) > 0:
print(f'\nCRITICAL: Found {len(duplicates)} duplicate username records!')
print(duplicates[['username', 'timestamp', 'na_count_psi_and_hemi']].head(20))
else:
print('\nPASS: No duplicate usernames found')
hash_duplicates = df[df.duplicated(subset=[uid_col], keep=False)]
if len(hash_duplicates) > 0:
print(f'\nCRITICAL: Found {len(hash_duplicates)} duplicate {uid_col} records!')
else:
print(f'PASS: All {uid_col} values are unique')
# --- Survey completeness ---
print_section('SURVEY COMPLETENESS')
na_dist = df['na_count_psi_and_hemi'].value_counts().sort_index()
complete_surveys = int((df['na_count_psi_and_hemi'] == 0).sum())
complete_pct = complete_surveys / len(df) * 100
print(f'Complete surveys (0 NAs): {complete_surveys:,} ({complete_pct:.1f}%)')
print(f'Partial surveys (1+ NAs): {len(df) - complete_surveys:,} ({100 - complete_pct:.1f}%)')
print(f'\nNA Count Distribution (0 = complete, 25 = all questions skipped):')
print(na_dist.head(10))
plt.figure(figsize=(12, 5))
na_dist.plot(kind='bar', color='coral')
plt.title('Survey Completeness Distribution', fontsize=14, fontweight='bold')
plt.xlabel('Number of Unanswered Questions')
plt.ylabel('Number of Users')
plt.tight_layout()
save_fig(output_dir, 'survey_completeness')
# Question-by-question completion
psi_completion = {col: (df[col].notna().sum() / len(df) * 100) for col in psi_cols}
hemi_completion = {col: (df[col].notna().sum() / len(df) * 100) for col in hemi_cols}
print('\nPSI QUESTION COMPLETION RATES')
for col, rate in psi_completion.items():
print(f'{col}: {rate:.1f}%')
print('\nHEMI QUESTION COMPLETION RATES')
for col, rate in hemi_completion.items():
print(f'{col}: {rate:.1f}%')
fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(16, 5))
ax1.bar(range(len(psi_completion)), list(psi_completion.values()), color='skyblue')
ax1.set_title('Psi Question Completion Rates', fontweight='bold')
ax1.set_xlabel('Question Number')
ax1.set_ylabel('Completion Rate (%)')
ax1.set_ylim(0, 100)
ax1.axhline(y=90, color='green', linestyle='--', alpha=0.5, label='90% threshold')
ax1.legend()
ax2.bar(range(len(hemi_completion)), list(hemi_completion.values()), color='lightcoral')
ax2.set_title('Hemi Question Completion Rates', fontweight='bold')
ax2.set_xlabel('Question Number')
ax2.set_ylabel('Completion Rate (%)')
ax2.set_ylim(0, 100)
ax2.axhline(y=90, color='green', linestyle='--', alpha=0.5, label='90% threshold')
ax2.legend()
plt.tight_layout()
save_fig(output_dir, 'question_completion_rates')
# --- File type distribution ---
if 'file_type' in df.columns:
print_section('FILE TYPE DISTRIBUTION')
file_type_dist = df['file_type'].value_counts()
print(file_type_dist)
plt.figure(figsize=(8, 5))
file_type_dist.plot(kind='bar', color=['steelblue', 'darkorange'])
plt.title('Source File Type Distribution', fontsize=14, fontweight='bold')
plt.xlabel('File Type')
plt.ylabel('Number of Records')
plt.xticks(rotation=0)
plt.tight_layout()
save_fig(output_dir, 'file_type_distribution')
print('\nFILE TYPE BY YEAR')
file_type_year = df.groupby(['year', 'file_type']).size().unstack(fill_value=0)
print(file_type_year)
file_type_year.plot(kind='bar', stacked=True, figsize=(14, 6))
plt.title('File Type Distribution by Year', fontsize=14, fontweight='bold')
plt.xlabel('Year')
plt.ylabel('Number of Records')
plt.legend(title='File Type')
plt.tight_layout()
save_fig(output_dir, 'file_type_by_year')
# --- Location data quality ---
print_section('LOCATION DATA QUALITY')
location_cols = ['state', 'coordinates', 'country']
has_any_location = 0
for col in location_cols:
if col in df.columns:
filled = df[col].notna().sum()
pct = filled / len(df) * 100
print(f'{col:12s}: {filled:6,} filled ({pct:5.1f}%)')
has_any_location = int(df[location_cols].notna().any(axis=1).sum())
has_all_location = int(df[location_cols].notna().all(axis=1).sum())
print(f'\nUsers with ANY location data: {has_any_location:,} ({has_any_location / len(df) * 100:.1f}%)')
print(f'Users with ALL location data: {has_all_location:,} ({has_all_location / len(df) * 100:.1f}%)')
if 'country' in df.columns and df['country'].notna().any():
print('\nTOP 20 COUNTRIES')
top_countries = df['country'].value_counts().head(20)
print(top_countries)
plt.figure(figsize=(12, 6))
top_countries.plot(kind='barh', color='teal')
plt.title('Top 20 Countries by User Count', fontsize=14, fontweight='bold')
plt.xlabel('Number of Users')
plt.ylabel('Country')
plt.gca().invert_yaxis()
plt.tight_layout()
save_fig(output_dir, 'top_countries')
# --- Validation ---
print_section('RESPONSE VALIDATION')
print('Psi/Hemi responses must be 1-5 or NULL')
all_survey_cols = psi_cols + hemi_cols
invalid_responses = {}
for col in all_survey_cols:
invalid = df[(df[col].notna()) & ((df[col] < 1) | (df[col] > 5))]
if len(invalid) > 0:
invalid_responses[col] = len(invalid)
if invalid_responses:
print('FAIL: Found invalid responses (not in 1-5 range):')
for col, count in invalid_responses.items():
print(f' {col}: {count} invalid values')
else:
print('PASS: All Psi/Hemi responses are in valid range (1-5)')
# na_count consistency
print('\nna_count_psi_and_hemi should match actual NAs')
actual_na_count = df[all_survey_cols].isnull().sum(axis=1)
reported_na_count = df['na_count_psi_and_hemi']
mismatch = (actual_na_count != reported_na_count)
mismatch_count = int(mismatch.sum())
if mismatch_count > 0:
print(f'WARNING: {mismatch_count} rows with na_count mismatch')
sample = df[mismatch][['username_hash', 'na_count_psi_and_hemi']].head(10).copy()
sample['actual_na_count'] = actual_na_count[mismatch].head(10).values
print(sample)
else:
print('PASS: All na_count_psi_and_hemi values match actual NA counts')
# --- Summary ---
standard_audit_summary('Users Dataset', df, extra_lines=[
f'Complete Surveys: {complete_surveys:,} ({complete_pct:.1f}%)',
f'Location Coverage: {has_any_location / len(df) * 100:.1f}%',
])
if __name__ == '__main__':
main()
|