-
Notifications
You must be signed in to change notification settings - Fork 0
/
collect_comparisons.py
271 lines (234 loc) · 10.4 KB
/
collect_comparisons.py
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
"""
This script runs through the "output folder"
to find all relevant comparisons
Then it formats these into a nice CSV for publication!
"""
import argparse
import json
import seaborn as sns
import matplotlib
import matplotlib.pyplot as plt
import pandas as pd
from data_helpers import strip_domain_strings_wrapper
from constants import FULL, TOP_THREE
from analysis import UGC_WHITELIST
sns.set(style="whitegrid", palette='muted', color_codes=True)
def parse():
"""
Collects comparisons
"""
parser = argparse.ArgumentParser(description='see file.')
parser.add_argument(
'--output_dirs', help='Where are the outputs', nargs='+', required=False)
parser.add_argument(
'--criteria', help='how are we identifying the significant comparisons', default='ttest')
parser.set_defaults(print_all=False)
args = parser.parse_args()
if args.output_dirs:
paths = args.output_dirs
else:
comparisons_and_queries = [
('urban-rural', 'all',),
('income', 'all'),
('voting', 'all'),
('2018-01-16_income', 'extra'),
('2018-01-17_voting', 'extra'),
('2018-01-17_urban-rural', 'extra'),
]
categories = [
'popular',
'procon_popular',
'trending',
'top_insurance',
'top_loans',
'med_sample_first_20',
]
paths = []
for cq in comparisons_and_queries:
for category in categories:
path = 'output/{comparison}_10_{queries}.db__{category}'.format(**{
'comparison': cq[0],
'queries': cq[1],
'category': category,
})
paths.append(path)
row_dicts = []
all_tests_together = None
ugc_row_dicts = []
n_ugc_diffs = 0
n_tests = 0
n_ugc_diffs_strict = 0
sub_to_domain = {
FULL: {},
TOP_THREE: {}
}
for path in paths:
pre, category = path.split('__')
comparison = pre.split('__')[0][7:]
print(path, '|', comparison, '|', category)
if args.criteria == 'ttest':
full = path + '/full_pval_summary.csv'
top3 = path + '/top_three_pval_summary.csv'
elif args.criteria == 'fisher':
full = path + '/full_fisher_summary.csv'
top3 = path + '/top_three_fisher_summary.csv'
all_tests = path + '/comparisons.csv'
try:
full_df = pd.read_csv(full)
top3_df = pd.read_csv(top3)
all_tests_df = pd.read_csv(all_tests)
except:
print('Not found: {}'.format(full))
continue
n_tests += len(all_tests_df.index)
all_tests_df['category'] = category
all_tests_df['comparison'] = comparison
if all_tests_together is None:
all_tests_together = all_tests_df
else:
all_tests_together = pd.concat([
all_tests_together, all_tests_df
])
strip_full = strip_domain_strings_wrapper(FULL)
strip_top3 = strip_domain_strings_wrapper(TOP_THREE)
for (df, subset, strip_func) in [
(full_df, FULL, strip_full),
(top3_df, TOP_THREE, strip_top3)
]:
for _, row in df.iterrows():
row_dict = {}
domain = strip_func(row['column'])
row_dict['domain'] = domain
row_dict['increase'] = round(row.add_inc, 2)
row_dict['winner'] = row.winner
row_dict['subset'] = subset
row_dict['category'] = category
row_dict['fisher_pval'] = row.fisher_pval
row_dicts.append(row_dict)
# row_dict['ratio'] = row.mult_inc
should_plot = (
domain == 'wikipedia.org' or
domain == 'MapsLocations' or
':' in domain # domain was coded
)
if should_plot:
code_index = domain.find(':')
if code_index != -1:
code = domain[domain.find(':')+1:]
print(code)
if code[:2] != 'tt':
print('Skipping code {} b/c not UGC'.format(code))
continue
else:
row_dict['domain'] = domain = domain.replace(':'+code, '')
if domain != 'MapsLocations':
n_ugc_diffs += 1
if row['fisher_pval'] < 0.05 / 115:
n_ugc_diffs_strict += 1
print(domain)
ugc_row_dicts.append(row_dict)
if domain not in sub_to_domain[subset]:
sub_to_domain[subset][domain] = {
'categories': [],
'locations': [],
'category_in_location': [],
'max': 0,
}
if round(row.add_inc, 2) > sub_to_domain[subset][domain]['max']:
sub_to_domain[subset][domain]['max'] = round(row.add_inc, 2)
if category not in sub_to_domain[subset][domain]['categories']:
sub_to_domain[subset][domain]['categories'].append(category)
if row.winner not in sub_to_domain[subset][domain]['locations']:
sub_to_domain[subset][domain]['locations'].append(row.winner)
tmp = category + ' in ' + row.winner
if tmp not in sub_to_domain[subset][domain]['category_in_location']:
sub_to_domain[subset][domain]['category_in_location'].append(tmp)
outdf = pd.DataFrame(row_dicts)
#print(outdf)
outdf.to_csv('collect_comparisons.csv')
all_tests_together.to_csv('all_tests_together.csv')
ugc_outdf = pd.DataFrame(ugc_row_dicts)
for subset, domain_to_vals in sub_to_domain.items():
for domain, vals in domain_to_vals.items():
vals['categories'] = ', '.join(vals['categories'])
vals['locations'] = ', '.join(vals['locations'])
vals['category_in_location'] = ', '.join(vals['category_in_location'])
max_full_df, max_top3_df = pd.DataFrame(sub_to_domain[FULL]).transpose(), pd.DataFrame(sub_to_domain[TOP_THREE]).transpose()
max_full_df.sort_values(['category_in_location']).to_csv('max_full_comparisons.csv')
max_top3_df.sort_values(['category_in_location']).to_csv('max_top3_comparisons.csv')
domains_plus_winners = [
str(x) + ' (' + str(y) + ')' for x, y in zip(
list(ugc_outdf.domain),
list(ugc_outdf.winner)
)
]
ugc_outdf = ugc_outdf.assign(domains_plus_winners=domains_plus_winners)
sorted_ugc_outdf = ugc_outdf.sort_values(['domain', 'category', 'winner'])
# visualize it
ugc_outdf.loc[ugc_outdf.category == 'top_insurance', 'category'] = 'insurance'
ugc_outdf.loc[ugc_outdf.category == 'top_loans', 'category'] = 'loans'
sorted_ugc_outdf[['category', 'domain', 'winner', 'subset', 'increase', 'fisher_pval']].to_csv('{}_comparisons.csv'.format(args.criteria), index=False)
with open('{}_comparison_deets.json'.format(args.criteria), 'w') as f:
json.dump({
'n_ugc_diffs': n_ugc_diffs,
'n_tests': n_tests,
'n_ugc_diffs_strict': n_ugc_diffs_strict,
}, f)
row1_df = ugc_outdf[(ugc_outdf.winner == 'urban') | (ugc_outdf.winner == 'DEM') | (ugc_outdf.winner == 'high-income')]
order1 = sorted(row1_df.domains_plus_winners.drop_duplicates())
row2_df = ugc_outdf[(ugc_outdf.winner == 'rural') | (ugc_outdf.winner == 'GOP') | (ugc_outdf.winner == 'low-income')]
order2 = sorted(row2_df.domains_plus_winners.drop_duplicates())
# ugc_outdf.loc[ugc_outdf.winner == 'high-income', 'winner'] = 'top SES'
# ugc_outdf.loc[ugc_outdf.winner == 'low-income', 'winner'] = 'bottom SES'
# ugc_outdf.loc[ugc_outdf.winner == 'GOP', 'winner'] = 'top GOP'
# ugc_outdf.loc[ugc_outdf.winner == 'DEM', 'winner'] = 'top DEM'
orders = [order1, order2]
sns.set_context("paper", rc={"font.size":10, "font.family": "Times New Roman", "axes.titlesize":10,"axes.labelsize":10})
fig, axes = plt.subplots(2, 2, figsize=(6.5, 3.5), gridspec_kw = {'height_ratios':[8, 3]}, dpi=300)
for rownum, row_df in enumerate([row1_df, row2_df]):
col1_df = row_df[row_df.subset == FULL]
col2_df = row_df[row_df.subset == TOP_THREE]
sns.barplot(x='increase', y='domains_plus_winners', hue='category', order=orders[rownum],
data=col1_df, ax=axes[rownum, 0], ci=None,)
sns.barplot(x='increase', y='domains_plus_winners', hue='category', order=orders[rownum],
data=col2_df, ax=axes[rownum, 1], ci=None,)
axes[rownum, 0].set(ylabel='', xlabel='')
axes[rownum, 0].legend(loc='lower right', frameon=True)
axes[rownum, 1].set(ylabel='', xlabel='', yticks=[])
axes[rownum, 1].legend(loc='lower right', frameon=True)
for colnum in [0, 1]:
ax = axes[rownum, colnum]
sns.despine(ax=ax, bottom=True, left=True)
# ax.set(xlabel='Increase in incidence rate')
ax.set(xlabel='')
ax.hlines([x + 0.5 for x in range(len(orders[rownum]))], *ax.get_xlim(), linestyle='--', color='lightgray')
title_template = 'Increases in incidence rate\nIn {locations_str}\nConsidering {subset_str}'
axes[0, 0].set_xlabel(
title_template.format(**{
'subset_str': 'full results',
'locations_str': 'urban/high-income/DEM areas',
})
)
axes[0, 1].set_xlabel(
title_template.format(**{
'subset_str': 'top three results',
'locations_str': 'urban/high-income/DEM areas',
})
)
axes[1, 0].set_xlabel(
title_template.format(**{
'subset_str': 'full results',
'locations_str': 'rural/low-income/GOP areas',
})
)
axes[1, 1].set_xlabel(
title_template.format(**{
'subset_str': 'top three results',
'locations_str': 'rural/low-income/GOP areas',
})
)
plt.tight_layout()
fig.savefig('comparisons.png')
plt.show()
if __name__ == '__main__':
parse()