-
Notifications
You must be signed in to change notification settings - Fork 2
/
donors-choose-config.yaml
316 lines (266 loc) · 9.19 KB
/
donors-choose-config.yaml
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
config_version: 'v8'
model_comment: 'quickstart test run'
random_seed: 1995
temporal_config:
# first date our feature data is good
feature_start_time: '2000-01-01'
feature_end_time: '2013-06-01'
# first date our label data is good
# donorschoose: as far back as we have good donation data
label_start_time: '2011-09-02'
label_end_time: '2013-06-01'
model_update_frequency: '1month'
# length of time defining a test set
test_durations: ['1month']
# defines how far back a training set reaches
max_training_histories: ['12month']
# we sample every day, since new projects are posted
# every day
training_as_of_date_frequencies: ['1day']
test_as_of_date_frequencies: ['1day']
# like our project timeout
training_label_timespans: ['4month']
test_label_timespans: ['4month']
cohort_config:
query: |
SELECT distinct(entity_id), date_posted as as_of_date
FROM optimized.projects
WHERE date_posted = '{as_of_date}'::date - interval '1day'
label_config:
query: |
WITH donation_totals AS
(SELECT projects.entity_id,
sum(case when donation_to_project is null then 0 else donation_to_project end) as total_donations,
total_asking_price AS total_price
FROM optimized.projects
LEFT JOIN optimized.donations ON (donations.entity_id = projects.entity_id
and donations.donation_timestamp < (projects.date_posted
+ interval '{label_timespan}'))
WHERE projects.date_posted = '{as_of_date}'::date - interval '1day'
GROUP BY projects.entity_id, projects.total_asking_price)
SELECT entity_id,
(total_donations < total_price)::int AS outcome
FROM optimized.projects
RIGHT JOIN donation_totals using(entity_id)
name: 'quickstart_label'
feature_aggregations:
-
prefix: 'project_features'
from_obj: 'optimized.projects'
knowledge_date_column: 'date_posted'
aggregates_imputation:
all:
type: 'zero'
categoricals_imputation:
all:
type: 'null_category'
categoricals:
-
column: 'school_metro'
metrics:
- 'sum'
choice_query: 'select distinct school_metro from optimized.projects'
-
column: 'resource_type'
metrics:
- 'sum'
choice_query: 'select distinct resource_type from optimized.projects'
-
column: 'poverty_level'
metrics:
- 'sum'
choice_query: 'select distinct left(poverty_level, 8) from optimized.projects'
-
column: 'grade_level'
metrics:
- 'sum'
choice_query: 'select distinct grade_level from optimized.projects'
-
column: 'teacher_prefix'
metrics:
- 'sum'
choice_query: 'select distinct teacher_prefix from optimized.projects'
-
column: 'school_state'
metrics:
- 'sum'
choice_query: 'select distinct school_state from optimized.projects'
aggregates:
-
quantity: 'total_asking_price'
metrics:
- 'sum'
# Since our time-aggregate features are precomputed, feature interval is
# irrelvant. We keep 'all' as a default.
intervals: ['all']
-
prefix: 'resources_features'
from_obj: 'optimized.resources'
knowledge_date_column: 'date_posted'
aggregates_imputation:
all:
type: 'zero'
aggregates:
-
quantity: 'item_unit_price'
metrics:
- 'sum'
-
quantity: 'item_quantity'
metrics:
- 'sum'
intervals: ['all']
-
prefix: 'essay_features'
from_obj: 'optimized.essays'
knowledge_date_column: 'date_posted'
aggregates_imputation:
all:
type: 'zero'
aggregates:
-
quantity: 'length(essay)::int'
metrics:
- 'sum'
intervals: ['all']
-
prefix: 'donation_features'
from_obj: 'optimized.time_series_features'
knowledge_date_column: 'date_posted'
aggregates_imputation:
all:
type: 'constant'
value: 0 # for testing
aggregates:
- # proportion of fully funded projects posted within the last year at the same district
quantity: 'district_funding_rate_1yr'
metrics:
- 'sum'
- # average donations per project posted within the last year at the same district
quantity: 'district_avg_donations_1yr'
metrics:
- 'sum'
- # proportion of fully funded projects posted within the last two years at the same district
quantity: 'district_funding_rate_2yr'
metrics:
- 'sum'
- # average donations per project posted within the two years at the same district
quantity: 'district_avg_donations_2yr'
metrics:
- 'sum'
# teachers
- # proportion of fully funded projects posted within the last year by the same teacher
quantity: 'teacher_funding_rate_1yr'
metrics:
- 'sum'
- # average donations per project posted within the last year by the same teacher
quantity: 'teacher_avg_donations_1yr'
metrics:
- 'sum'
- # proportion of fully funded projects posted within the last two years by the same teacher
quantity: 'teacher_funding_rate_2yr'
metrics:
- 'sum'
- # average donations per project posted within the two years by the same teacher
quantity: 'teacher_avg_donations_2yr'
metrics:
- 'sum'
# zip
- # proportion of fully funded projects posted within the last year in the same zip code
quantity: 'zip_funding_rate_1yr'
metrics:
- 'sum'
- # average donations per project posted within the last year in the same zip code
quantity: 'zip_avg_donations_1yr'
metrics:
- 'sum'
- # proportion of fully funded projects posted within the last two years in the same zip code
quantity: 'zip_funding_rate_2yr'
metrics:
- 'sum'
- # average donations per project posted within the two years in the same zip code
quantity: 'zip_avg_donations_2yr'
metrics:
- 'sum'
intervals: ['all']
grid_config:
'sklearn.ensemble.RandomForestClassifier':
n_estimators: [1000, 2500, 5000, 10000]
max_depth: [2, 5, 10, 50]
max_features: ['sqrt', 'auto', 'log2']
min_samples_split: [2, 10, 25, 50]
'triage.component.catwalk.estimators.classifiers.ScaledLogisticRegression':
max_iter: [10000]
penalty: ['l1', 'l2']
C: [0.001, 0.01, 0.1, 0.5, 1, 2, 10]
solver: ['saga']
'sklearn.tree.DecisionTreeClassifier':
max_depth: [2, 5, 10, 50]
min_samples_split: [2, 10, 25, 50]
'sklearn.dummy.DummyClassifier':
# baseline, predicts base rate
strategy: ['prior']
'triage.component.catwalk.baselines.rankers.BaselineRankMultiFeature':
rules:
- [{feature: 'project_features_entity_id_all_total_asking_price_sum', low_value_high_score: False}]
scoring:
testing_metric_groups:
-
metrics: [precision@, recall@]
thresholds:
percentiles: [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]
top_n: [50, 100, 200, 500, 1000]
-
metrics: [roc_auc]
training_metric_groups:
-
metrics: [precision@, recall@]
thresholds:
percentiles: [1, 5, 10, 20, 30, 40, 50, 60, 70, 80, 90, 100]
subsets:
-
name: 'poor'
query: |
select
entity_id
from optimized.projects
where poverty_level in ('high_poverty', 'highest poverty')
and date_posted < '{as_of_date}'::date
-
name: 'urban'
query: |
select
entity_id
from optimized.projects
where school_metro = 'urban'
and date_posted < '{as_of_date}'::date
-
name: 'rural'
query: |
select
entity_id
from optimized.projects
where school_metro = 'rural'
and date_posted < '{as_of_date}'::date
bias_audit_config:
from_obj_table: 'optimized.projects'
attribute_columns:
- 'teacher_prefix'
knowledge_date_column: 'date_posted'
entity_id_column: 'entity_id'
ref_groups_method: 'predefined'
ref_groups:
'teacher_prefix': 'Mr.'
thresholds:
percentiles: [1, 5, 10, 15, 20, 25, 50, 100]
top_n: [50, 100, 200, 500, 1000]