lecrapaud 0.14.5__py3-none-any.whl → 0.14.6__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Potentially problematic release.

This version of lecrapaud might be problematic. Click here for more details.

Files changed (13) hide show
  1. lecrapaud/db/alembic/versions/2025_06_23_1748-f089dfb7e3ba_.py +423 -188
  2. lecrapaud/db/alembic/versions/2025_06_24_1216-c62251b129ed_.py +9 -4
  3. lecrapaud/db/alembic/versions/2025_06_24_1711-86457e2f333f_.py +19 -8
  4. lecrapaud/db/alembic/versions/2025_06_25_1759-72aa496ca65b_.py +9 -4
  5. lecrapaud/db/models/experiment.py +7 -16
  6. lecrapaud/db/models/feature_selection.py +7 -18
  7. lecrapaud/db/models/utils.py +32 -0
  8. lecrapaud/feature_engineering.py +6 -3
  9. lecrapaud/model_selection.py +7 -5
  10. {lecrapaud-0.14.5.dist-info → lecrapaud-0.14.6.dist-info}/METADATA +1 -1
  11. {lecrapaud-0.14.5.dist-info → lecrapaud-0.14.6.dist-info}/RECORD +13 -12
  12. {lecrapaud-0.14.5.dist-info → lecrapaud-0.14.6.dist-info}/LICENSE +0 -0
  13. {lecrapaud-0.14.5.dist-info → lecrapaud-0.14.6.dist-info}/WHEEL +0 -0
lecrapaud/db/alembic/versions/2025_06_23_1748-f089dfb7e3ba_.py CHANGED
@@ -1,18 +1,20 @@
1
1
  """
2
2
 
3
3
  Revision ID: f089dfb7e3ba
4
- Revises:
4
+ Revises:
5
5
  Create Date: 2025-06-23 17:48:32.842030
6
6
 
7
7
  """
8
+
8
9
  from typing import Sequence, Union
9
10
 
10
11
  from alembic import op
11
12
  import sqlalchemy as sa
13
+ from lecrapaud.config import LECRAPAUD_TABLE_PREFIX
12
14
 
13
15
 
14
16
  # revision identifiers, used by Alembic.
15
- revision: str = 'f089dfb7e3ba'
17
+ revision: str = "f089dfb7e3ba"
16
18
  down_revision: Union[str, None] = None
17
19
  branch_labels: Union[str, Sequence[str], None] = None
18
20
  depends_on: Union[str, Sequence[str], None] = None
@@ -20,195 +22,428 @@ depends_on: Union[str, Sequence[str], None] = None
20
22
 
21
23
  def upgrade() -> None:
22
24
  # ### commands auto generated by Alembic - please adjust! ###
23
- op.create_table('lecrapaud_experiments',
24
- sa.Column('id', sa.BigInteger(), autoincrement=True, nullable=False),
25
- sa.Column('created_at', sa.TIMESTAMP(timezone=True), server_default=sa.text('now()'), nullable=False),
26
- sa.Column('updated_at', sa.TIMESTAMP(timezone=True), server_default=sa.text('now()'), nullable=False),
27
- sa.Column('name', sa.String(length=50), nullable=False),
28
- sa.Column('path', sa.String(length=255), nullable=True),
29
- sa.Column('type', sa.String(length=50), nullable=False),
30
- sa.Column('size', sa.Integer(), nullable=False),
31
- sa.Column('train_size', sa.Integer(), nullable=True),
32
- sa.Column('val_size', sa.Integer(), nullable=True),
33
- sa.Column('test_size', sa.Integer(), nullable=True),
34
- sa.Column('corr_threshold', sa.Float(), nullable=False),
35
- sa.Column('max_features', sa.Integer(), nullable=False),
36
- sa.Column('percentile', sa.Float(), nullable=False),
37
- sa.Column('number_of_groups', sa.Integer(), nullable=True),
38
- sa.Column('list_of_groups', sa.JSON(), nullable=True),
39
- sa.Column('start_date', sa.DateTime(), nullable=True),
40
- sa.Column('end_date', sa.DateTime(), nullable=True),
41
- sa.Column('train_start_date', sa.DateTime(), nullable=True),
42
- sa.Column('train_end_date', sa.DateTime(), nullable=True),
43
- sa.Column('val_start_date', sa.DateTime(), nullable=True),
44
- sa.Column('val_end_date', sa.DateTime(), nullable=True),
45
- sa.Column('test_start_date', sa.DateTime(), nullable=True),
46
- sa.Column('test_end_date', sa.DateTime(), nullable=True),
47
- sa.PrimaryKeyConstraint('id'),
48
- sa.UniqueConstraint('name', name='uq_experiments_composite')
49
- )
50
- op.create_index(op.f('ix_lecrapaud_experiments_id'), 'lecrapaud_experiments', ['id'], unique=False)
51
- op.create_table('lecrapaud_features',
52
- sa.Column('id', sa.BigInteger(), autoincrement=True, nullable=False),
53
- sa.Column('created_at', sa.TIMESTAMP(timezone=True), server_default=sa.text('now()'), nullable=False),
54
- sa.Column('updated_at', sa.TIMESTAMP(timezone=True), server_default=sa.text('now()'), nullable=False),
55
- sa.Column('name', sa.String(length=50), nullable=False),
56
- sa.Column('type', sa.String(length=50), nullable=True),
57
- sa.PrimaryKeyConstraint('id'),
58
- sa.UniqueConstraint('name')
59
- )
60
- op.create_index(op.f('ix_lecrapaud_features_id'), 'lecrapaud_features', ['id'], unique=False)
61
- op.create_table('lecrapaud_models',
62
- sa.Column('id', sa.BigInteger(), autoincrement=True, nullable=False),
63
- sa.Column('created_at', sa.TIMESTAMP(timezone=True), server_default=sa.text('now()'), nullable=False),
64
- sa.Column('updated_at', sa.TIMESTAMP(timezone=True), server_default=sa.text('now()'), nullable=False),
65
- sa.Column('name', sa.String(length=50), nullable=False),
66
- sa.Column('type', sa.String(length=50), nullable=False),
67
- sa.PrimaryKeyConstraint('id'),
68
- sa.UniqueConstraint('name', 'type', name='uq_model_composite')
69
- )
70
- op.create_index(op.f('ix_lecrapaud_models_id'), 'lecrapaud_models', ['id'], unique=False)
71
- op.create_table('lecrapaud_targets',
72
- sa.Column('id', sa.BigInteger(), autoincrement=True, nullable=False),
73
- sa.Column('created_at', sa.TIMESTAMP(timezone=True), server_default=sa.text('now()'), nullable=False),
74
- sa.Column('updated_at', sa.TIMESTAMP(timezone=True), server_default=sa.text('now()'), nullable=False),
75
- sa.Column('name', sa.String(length=50), nullable=False),
76
- sa.Column('type', sa.String(length=50), nullable=False),
77
- sa.Column('description', sa.String(length=255), nullable=True),
78
- sa.PrimaryKeyConstraint('id'),
79
- sa.UniqueConstraint('name', 'type', name='uq_target_composite')
80
- )
81
- op.create_index(op.f('ix_lecrapaud_targets_id'), 'lecrapaud_targets', ['id'], unique=False)
82
- op.create_table('lecrapaud_experiment_target_association',
83
- sa.Column('experiment_id', sa.BigInteger(), nullable=False),
84
- sa.Column('target_id', sa.BigInteger(), nullable=False),
85
- sa.ForeignKeyConstraint(['experiment_id'], ['lecrapaud_experiments.id'], ondelete='CASCADE'),
86
- sa.ForeignKeyConstraint(['target_id'], ['lecrapaud_targets.id'], ondelete='CASCADE'),
87
- sa.PrimaryKeyConstraint('experiment_id', 'target_id')
88
- )
89
- op.create_table('lecrapaud_feature_selections',
90
- sa.Column('id', sa.BigInteger(), autoincrement=True, nullable=False),
91
- sa.Column('created_at', sa.TIMESTAMP(timezone=True), server_default=sa.text('now()'), nullable=False),
92
- sa.Column('updated_at', sa.TIMESTAMP(timezone=True), server_default=sa.text('now()'), nullable=False),
93
- sa.Column('training_time', sa.Integer(), nullable=True),
94
- sa.Column('best_features_path', sa.String(length=255), nullable=True),
95
- sa.Column('experiment_id', sa.BigInteger(), nullable=False),
96
- sa.Column('target_id', sa.BigInteger(), nullable=False),
97
- sa.ForeignKeyConstraint(['experiment_id'], ['lecrapaud_experiments.id'], ondelete='CASCADE'),
98
- sa.ForeignKeyConstraint(['target_id'], ['lecrapaud_targets.id'], ondelete='CASCADE'),
99
- sa.PrimaryKeyConstraint('id'),
100
- sa.UniqueConstraint('experiment_id', 'target_id', name='uq_feature_selection_composite')
101
- )
102
- op.create_index(op.f('ix_lecrapaud_feature_selections_id'), 'lecrapaud_feature_selections', ['id'], unique=False)
103
- op.create_table('lecrapaud_model_selections',
104
- sa.Column('id', sa.BigInteger(), autoincrement=True, nullable=False),
105
- sa.Column('created_at', sa.TIMESTAMP(timezone=True), server_default=sa.text('now()'), nullable=False),
106
- sa.Column('updated_at', sa.TIMESTAMP(timezone=True), server_default=sa.text('now()'), nullable=False),
107
- sa.Column('best_model_params', sa.JSON(), nullable=True),
108
- sa.Column('best_model_path', sa.String(length=255), nullable=True),
109
- sa.Column('best_model_id', sa.BigInteger(), nullable=True),
110
- sa.Column('target_id', sa.BigInteger(), nullable=False),
111
- sa.Column('experiment_id', sa.BigInteger(), nullable=False),
112
- sa.ForeignKeyConstraint(['best_model_id'], ['lecrapaud_models.id'], ondelete='CASCADE'),
113
- sa.ForeignKeyConstraint(['experiment_id'], ['lecrapaud_experiments.id'], ondelete='CASCADE'),
114
- sa.ForeignKeyConstraint(['target_id'], ['lecrapaud_targets.id'], ondelete='CASCADE'),
115
- sa.PrimaryKeyConstraint('id'),
116
- sa.UniqueConstraint('target_id', 'experiment_id', name='uq_model_selection_composite')
117
- )
118
- op.create_index(op.f('ix_lecrapaud_model_selections_id'), 'lecrapaud_model_selections', ['id'], unique=False)
119
- op.create_table('lecrapaud_feature_selection_association',
120
- sa.Column('feature_selection_id', sa.BigInteger(), nullable=False),
121
- sa.Column('feature_id', sa.BigInteger(), nullable=False),
122
- sa.ForeignKeyConstraint(['feature_id'], ['lecrapaud_features.id'], ondelete='CASCADE'),
123
- sa.ForeignKeyConstraint(['feature_selection_id'], ['lecrapaud_feature_selections.id'], ondelete='CASCADE'),
124
- sa.PrimaryKeyConstraint('feature_selection_id', 'feature_id')
125
- )
126
- op.create_table('lecrapaud_feature_selection_ranks',
127
- sa.Column('id', sa.BigInteger(), autoincrement=True, nullable=False),
128
- sa.Column('created_at', sa.TIMESTAMP(timezone=True), server_default=sa.text('now()'), nullable=False),
129
- sa.Column('updated_at', sa.TIMESTAMP(timezone=True), server_default=sa.text('now()'), nullable=False),
130
- sa.Column('score', sa.Float(), nullable=True),
131
- sa.Column('pvalue', sa.Float(), nullable=True),
132
- sa.Column('support', sa.Integer(), nullable=True),
133
- sa.Column('rank', sa.Integer(), nullable=True),
134
- sa.Column('method', sa.String(length=50), nullable=True),
135
- sa.Column('training_time', sa.Integer(), nullable=True),
136
- sa.Column('feature_id', sa.BigInteger(), nullable=True),
137
- sa.Column('feature_selection_id', sa.BigInteger(), nullable=True),
138
- sa.ForeignKeyConstraint(['feature_id'], ['lecrapaud_features.id'], ondelete='CASCADE'),
139
- sa.ForeignKeyConstraint(['feature_selection_id'], ['lecrapaud_feature_selections.id'], ondelete='CASCADE'),
140
- sa.PrimaryKeyConstraint('id'),
141
- sa.UniqueConstraint('feature_id', 'feature_selection_id', 'method', name='uq_feature_selection_rank_composite')
142
- )
143
- op.create_index(op.f('ix_lecrapaud_feature_selection_ranks_id'), 'lecrapaud_feature_selection_ranks', ['id'], unique=False)
144
- op.create_table('lecrapaud_model_trainings',
145
- sa.Column('id', sa.BigInteger(), autoincrement=True, nullable=False),
146
- sa.Column('created_at', sa.TIMESTAMP(timezone=True), server_default=sa.text('now()'), nullable=False),
147
- sa.Column('updated_at', sa.TIMESTAMP(timezone=True), server_default=sa.text('now()'), nullable=False),
148
- sa.Column('best_params', sa.JSON(), nullable=True),
149
- sa.Column('model_path', sa.String(length=255), nullable=True),
150
- sa.Column('training_time', sa.Integer(), nullable=True),
151
- sa.Column('model_id', sa.BigInteger(), nullable=False),
152
- sa.Column('model_selection_id', sa.BigInteger(), nullable=False),
153
- sa.ForeignKeyConstraint(['model_id'], ['lecrapaud_models.id'], ),
154
- sa.ForeignKeyConstraint(['model_selection_id'], ['lecrapaud_model_selections.id'], ondelete='CASCADE'),
155
- sa.PrimaryKeyConstraint('id'),
156
- sa.UniqueConstraint('model_id', 'model_selection_id', name='uq_model_training_composite')
157
- )
158
- op.create_index(op.f('ix_lecrapaud_model_trainings_id'), 'lecrapaud_model_trainings', ['id'], unique=False)
159
- op.create_table('lecrapaud_scores',
160
- sa.Column('id', sa.BigInteger(), autoincrement=True, nullable=False),
161
- sa.Column('created_at', sa.TIMESTAMP(timezone=True), server_default=sa.text('now()'), nullable=False),
162
- sa.Column('updated_at', sa.TIMESTAMP(timezone=True), server_default=sa.text('now()'), nullable=False),
163
- sa.Column('type', sa.String(length=50), nullable=False),
164
- sa.Column('training_time', sa.Integer(), nullable=True),
165
- sa.Column('eval_data_std', sa.Float(), nullable=True),
166
- sa.Column('rmse', sa.Float(), nullable=True),
167
- sa.Column('rmse_std_ratio', sa.Float(), nullable=True),
168
- sa.Column('mae', sa.Float(), nullable=True),
169
- sa.Column('mape', sa.Float(), nullable=True),
170
- sa.Column('mam', sa.Float(), nullable=True),
171
- sa.Column('mad', sa.Float(), nullable=True),
172
- sa.Column('mae_mam_ratio', sa.Float(), nullable=True),
173
- sa.Column('mae_mad_ratio', sa.Float(), nullable=True),
174
- sa.Column('r2', sa.Float(), nullable=True),
175
- sa.Column('logloss', sa.Float(), nullable=True),
176
- sa.Column('accuracy', sa.Float(), nullable=True),
177
- sa.Column('precision', sa.Float(), nullable=True),
178
- sa.Column('recall', sa.Float(), nullable=True),
179
- sa.Column('f1', sa.Float(), nullable=True),
180
- sa.Column('roc_auc', sa.Float(), nullable=True),
181
- sa.Column('threshold', sa.Float(), nullable=True),
182
- sa.Column('precision_at_threshold', sa.Float(), nullable=True),
183
- sa.Column('recall_at_threshold', sa.Float(), nullable=True),
184
- sa.Column('model_training_id', sa.BigInteger(), nullable=False),
185
- sa.ForeignKeyConstraint(['model_training_id'], ['lecrapaud_model_trainings.id'], ondelete='CASCADE'),
186
- sa.PrimaryKeyConstraint('id')
187
- )
188
- op.create_index(op.f('ix_lecrapaud_scores_id'), 'lecrapaud_scores', ['id'], unique=False)
25
+ op.create_table(
26
+ f"{LECRAPAUD_TABLE_PREFIX}_experiments",
27
+ sa.Column("id", sa.BigInteger(), autoincrement=True, nullable=False),
28
+ sa.Column(
29
+ "created_at",
30
+ sa.TIMESTAMP(timezone=True),
31
+ server_default=sa.text("now()"),
32
+ nullable=False,
33
+ ),
34
+ sa.Column(
35
+ "updated_at",
36
+ sa.TIMESTAMP(timezone=True),
37
+ server_default=sa.text("now()"),
38
+ nullable=False,
39
+ ),
40
+ sa.Column("name", sa.String(length=50), nullable=False),
41
+ sa.Column("path", sa.String(length=255), nullable=True),
42
+ sa.Column("type", sa.String(length=50), nullable=False),
43
+ sa.Column("size", sa.Integer(), nullable=False),
44
+ sa.Column("train_size", sa.Integer(), nullable=True),
45
+ sa.Column("val_size", sa.Integer(), nullable=True),
46
+ sa.Column("test_size", sa.Integer(), nullable=True),
47
+ sa.Column("corr_threshold", sa.Float(), nullable=False),
48
+ sa.Column("max_features", sa.Integer(), nullable=False),
49
+ sa.Column("percentile", sa.Float(), nullable=False),
50
+ sa.Column("number_of_groups", sa.Integer(), nullable=True),
51
+ sa.Column("list_of_groups", sa.JSON(), nullable=True),
52
+ sa.Column("start_date", sa.DateTime(), nullable=True),
53
+ sa.Column("end_date", sa.DateTime(), nullable=True),
54
+ sa.Column("train_start_date", sa.DateTime(), nullable=True),
55
+ sa.Column("train_end_date", sa.DateTime(), nullable=True),
56
+ sa.Column("val_start_date", sa.DateTime(), nullable=True),
57
+ sa.Column("val_end_date", sa.DateTime(), nullable=True),
58
+ sa.Column("test_start_date", sa.DateTime(), nullable=True),
59
+ sa.Column("test_end_date", sa.DateTime(), nullable=True),
60
+ sa.PrimaryKeyConstraint("id"),
61
+ sa.UniqueConstraint("name", name="uq_experiments_composite"),
62
+ )
63
+ op.create_index(
64
+ op.f(f"ix_{LECRAPAUD_TABLE_PREFIX}_experiments_id"),
65
+ f"{LECRAPAUD_TABLE_PREFIX}_experiments",
66
+ ["id"],
67
+ unique=False,
68
+ )
69
+ op.create_table(
70
+ f"{LECRAPAUD_TABLE_PREFIX}_features",
71
+ sa.Column("id", sa.BigInteger(), autoincrement=True, nullable=False),
72
+ sa.Column(
73
+ "created_at",
74
+ sa.TIMESTAMP(timezone=True),
75
+ server_default=sa.text("now()"),
76
+ nullable=False,
77
+ ),
78
+ sa.Column(
79
+ "updated_at",
80
+ sa.TIMESTAMP(timezone=True),
81
+ server_default=sa.text("now()"),
82
+ nullable=False,
83
+ ),
84
+ sa.Column("name", sa.String(length=50), nullable=False),
85
+ sa.Column("type", sa.String(length=50), nullable=True),
86
+ sa.PrimaryKeyConstraint("id"),
87
+ sa.UniqueConstraint("name"),
88
+ )
89
+ op.create_index(
90
+ op.f(f"ix_{LECRAPAUD_TABLE_PREFIX}_features_id"),
91
+ f"{LECRAPAUD_TABLE_PREFIX}_features",
92
+ ["id"],
93
+ unique=False,
94
+ )
95
+ op.create_table(
96
+ f"{LECRAPAUD_TABLE_PREFIX}_models",
97
+ sa.Column("id", sa.BigInteger(), autoincrement=True, nullable=False),
98
+ sa.Column(
99
+ "created_at",
100
+ sa.TIMESTAMP(timezone=True),
101
+ server_default=sa.text("now()"),
102
+ nullable=False,
103
+ ),
104
+ sa.Column(
105
+ "updated_at",
106
+ sa.TIMESTAMP(timezone=True),
107
+ server_default=sa.text("now()"),
108
+ nullable=False,
109
+ ),
110
+ sa.Column("name", sa.String(length=50), nullable=False),
111
+ sa.Column("type", sa.String(length=50), nullable=False),
112
+ sa.PrimaryKeyConstraint("id"),
113
+ sa.UniqueConstraint("name", "type", name="uq_model_composite"),
114
+ )
115
+ op.create_index(
116
+ op.f(f"ix_{LECRAPAUD_TABLE_PREFIX}_models_id"),
117
+ f"{LECRAPAUD_TABLE_PREFIX}_models",
118
+ ["id"],
119
+ unique=False,
120
+ )
121
+ op.create_table(
122
+ f"{LECRAPAUD_TABLE_PREFIX}_targets",
123
+ sa.Column("id", sa.BigInteger(), autoincrement=True, nullable=False),
124
+ sa.Column(
125
+ "created_at",
126
+ sa.TIMESTAMP(timezone=True),
127
+ server_default=sa.text("now()"),
128
+ nullable=False,
129
+ ),
130
+ sa.Column(
131
+ "updated_at",
132
+ sa.TIMESTAMP(timezone=True),
133
+ server_default=sa.text("now()"),
134
+ nullable=False,
135
+ ),
136
+ sa.Column("name", sa.String(length=50), nullable=False),
137
+ sa.Column("type", sa.String(length=50), nullable=False),
138
+ sa.Column("description", sa.String(length=255), nullable=True),
139
+ sa.PrimaryKeyConstraint("id"),
140
+ sa.UniqueConstraint("name", "type", name="uq_target_composite"),
141
+ )
142
+ op.create_index(
143
+ op.f(f"ix_{LECRAPAUD_TABLE_PREFIX}_targets_id"),
144
+ f"{LECRAPAUD_TABLE_PREFIX}_targets",
145
+ ["id"],
146
+ unique=False,
147
+ )
148
+ op.create_table(
149
+ f"{LECRAPAUD_TABLE_PREFIX}_experiment_target_association",
150
+ sa.Column("experiment_id", sa.BigInteger(), nullable=False),
151
+ sa.Column("target_id", sa.BigInteger(), nullable=False),
152
+ sa.ForeignKeyConstraint(
153
+ ["experiment_id"],
154
+ [f"{LECRAPAUD_TABLE_PREFIX}_experiments.id"],
155
+ ondelete="CASCADE",
156
+ ),
157
+ sa.ForeignKeyConstraint(
158
+ ["target_id"], [f"{LECRAPAUD_TABLE_PREFIX}_targets.id"], ondelete="CASCADE"
159
+ ),
160
+ sa.PrimaryKeyConstraint("experiment_id", "target_id"),
161
+ )
162
+ op.create_table(
163
+ f"{LECRAPAUD_TABLE_PREFIX}_feature_selections",
164
+ sa.Column("id", sa.BigInteger(), autoincrement=True, nullable=False),
165
+ sa.Column(
166
+ "created_at",
167
+ sa.TIMESTAMP(timezone=True),
168
+ server_default=sa.text("now()"),
169
+ nullable=False,
170
+ ),
171
+ sa.Column(
172
+ "updated_at",
173
+ sa.TIMESTAMP(timezone=True),
174
+ server_default=sa.text("now()"),
175
+ nullable=False,
176
+ ),
177
+ sa.Column("training_time", sa.Integer(), nullable=True),
178
+ sa.Column("best_features_path", sa.String(length=255), nullable=True),
179
+ sa.Column("experiment_id", sa.BigInteger(), nullable=False),
180
+ sa.Column("target_id", sa.BigInteger(), nullable=False),
181
+ sa.ForeignKeyConstraint(
182
+ ["experiment_id"],
183
+ [f"{LECRAPAUD_TABLE_PREFIX}_experiments.id"],
184
+ ondelete="CASCADE",
185
+ ),
186
+ sa.ForeignKeyConstraint(
187
+ ["target_id"], [f"{LECRAPAUD_TABLE_PREFIX}_targets.id"], ondelete="CASCADE"
188
+ ),
189
+ sa.PrimaryKeyConstraint("id"),
190
+ sa.UniqueConstraint(
191
+ "experiment_id", "target_id", name="uq_feature_selection_composite"
192
+ ),
193
+ )
194
+ op.create_index(
195
+ op.f(f"ix_{LECRAPAUD_TABLE_PREFIX}_feature_selections_id"),
196
+ f"{LECRAPAUD_TABLE_PREFIX}_feature_selections",
197
+ ["id"],
198
+ unique=False,
199
+ )
200
+ op.create_table(
201
+ f"{LECRAPAUD_TABLE_PREFIX}_model_selections",
202
+ sa.Column("id", sa.BigInteger(), autoincrement=True, nullable=False),
203
+ sa.Column(
204
+ "created_at",
205
+ sa.TIMESTAMP(timezone=True),
206
+ server_default=sa.text("now()"),
207
+ nullable=False,
208
+ ),
209
+ sa.Column(
210
+ "updated_at",
211
+ sa.TIMESTAMP(timezone=True),
212
+ server_default=sa.text("now()"),
213
+ nullable=False,
214
+ ),
215
+ sa.Column("best_model_params", sa.JSON(), nullable=True),
216
+ sa.Column("best_model_path", sa.String(length=255), nullable=True),
217
+ sa.Column("best_model_id", sa.BigInteger(), nullable=True),
218
+ sa.Column("target_id", sa.BigInteger(), nullable=False),
219
+ sa.Column("experiment_id", sa.BigInteger(), nullable=False),
220
+ sa.ForeignKeyConstraint(
221
+ ["best_model_id"],
222
+ [f"{LECRAPAUD_TABLE_PREFIX}_models.id"],
223
+ ondelete="CASCADE",
224
+ ),
225
+ sa.ForeignKeyConstraint(
226
+ ["experiment_id"],
227
+ [f"{LECRAPAUD_TABLE_PREFIX}_experiments.id"],
228
+ ondelete="CASCADE",
229
+ ),
230
+ sa.ForeignKeyConstraint(
231
+ ["target_id"], [f"{LECRAPAUD_TABLE_PREFIX}_targets.id"], ondelete="CASCADE"
232
+ ),
233
+ sa.PrimaryKeyConstraint("id"),
234
+ sa.UniqueConstraint(
235
+ "target_id", "experiment_id", name="uq_model_selection_composite"
236
+ ),
237
+ )
238
+ op.create_index(
239
+ op.f(f"ix_{LECRAPAUD_TABLE_PREFIX}_model_selections_id"),
240
+ f"{LECRAPAUD_TABLE_PREFIX}_model_selections",
241
+ ["id"],
242
+ unique=False,
243
+ )
244
+ op.create_table(
245
+ f"{LECRAPAUD_TABLE_PREFIX}_feature_selection_association",
246
+ sa.Column("feature_selection_id", sa.BigInteger(), nullable=False),
247
+ sa.Column("feature_id", sa.BigInteger(), nullable=False),
248
+ sa.ForeignKeyConstraint(
249
+ ["feature_id"],
250
+ [f"{LECRAPAUD_TABLE_PREFIX}_features.id"],
251
+ ondelete="CASCADE",
252
+ ),
253
+ sa.ForeignKeyConstraint(
254
+ ["feature_selection_id"],
255
+ [f"{LECRAPAUD_TABLE_PREFIX}_feature_selections.id"],
256
+ ondelete="CASCADE",
257
+ ),
258
+ sa.PrimaryKeyConstraint("feature_selection_id", "feature_id"),
259
+ )
260
+ op.create_table(
261
+ f"{LECRAPAUD_TABLE_PREFIX}_feature_selection_ranks",
262
+ sa.Column("id", sa.BigInteger(), autoincrement=True, nullable=False),
263
+ sa.Column(
264
+ "created_at",
265
+ sa.TIMESTAMP(timezone=True),
266
+ server_default=sa.text("now()"),
267
+ nullable=False,
268
+ ),
269
+ sa.Column(
270
+ "updated_at",
271
+ sa.TIMESTAMP(timezone=True),
272
+ server_default=sa.text("now()"),
273
+ nullable=False,
274
+ ),
275
+ sa.Column("score", sa.Float(), nullable=True),
276
+ sa.Column("pvalue", sa.Float(), nullable=True),
277
+ sa.Column("support", sa.Integer(), nullable=True),
278
+ sa.Column("rank", sa.Integer(), nullable=True),
279
+ sa.Column("method", sa.String(length=50), nullable=True),
280
+ sa.Column("training_time", sa.Integer(), nullable=True),
281
+ sa.Column("feature_id", sa.BigInteger(), nullable=True),
282
+ sa.Column("feature_selection_id", sa.BigInteger(), nullable=True),
283
+ sa.ForeignKeyConstraint(
284
+ ["feature_id"],
285
+ [f"{LECRAPAUD_TABLE_PREFIX}_features.id"],
286
+ ondelete="CASCADE",
287
+ ),
288
+ sa.ForeignKeyConstraint(
289
+ ["feature_selection_id"],
290
+ [f"{LECRAPAUD_TABLE_PREFIX}_feature_selections.id"],
291
+ ondelete="CASCADE",
292
+ ),
293
+ sa.PrimaryKeyConstraint("id"),
294
+ sa.UniqueConstraint(
295
+ "feature_id",
296
+ "feature_selection_id",
297
+ "method",
298
+ name="uq_feature_selection_rank_composite",
299
+ ),
300
+ )
301
+ op.create_index(
302
+ op.f(f"ix_{LECRAPAUD_TABLE_PREFIX}_feature_selection_ranks_id"),
303
+ f"{LECRAPAUD_TABLE_PREFIX}_feature_selection_ranks",
304
+ ["id"],
305
+ unique=False,
306
+ )
307
+ op.create_table(
308
+ f"{LECRAPAUD_TABLE_PREFIX}_model_trainings",
309
+ sa.Column("id", sa.BigInteger(), autoincrement=True, nullable=False),
310
+ sa.Column(
311
+ "created_at",
312
+ sa.TIMESTAMP(timezone=True),
313
+ server_default=sa.text("now()"),
314
+ nullable=False,
315
+ ),
316
+ sa.Column(
317
+ "updated_at",
318
+ sa.TIMESTAMP(timezone=True),
319
+ server_default=sa.text("now()"),
320
+ nullable=False,
321
+ ),
322
+ sa.Column("best_params", sa.JSON(), nullable=True),
323
+ sa.Column("model_path", sa.String(length=255), nullable=True),
324
+ sa.Column("training_time", sa.Integer(), nullable=True),
325
+ sa.Column("model_id", sa.BigInteger(), nullable=False),
326
+ sa.Column("model_selection_id", sa.BigInteger(), nullable=False),
327
+ sa.ForeignKeyConstraint(
328
+ ["model_id"],
329
+ [f"{LECRAPAUD_TABLE_PREFIX}_models.id"],
330
+ ),
331
+ sa.ForeignKeyConstraint(
332
+ ["model_selection_id"],
333
+ [f"{LECRAPAUD_TABLE_PREFIX}_model_selections.id"],
334
+ ondelete="CASCADE",
335
+ ),
336
+ sa.PrimaryKeyConstraint("id"),
337
+ sa.UniqueConstraint(
338
+ "model_id", "model_selection_id", name="uq_model_training_composite"
339
+ ),
340
+ )
341
+ op.create_index(
342
+ op.f(f"ix_{LECRAPAUD_TABLE_PREFIX}_model_trainings_id"),
343
+ f"{LECRAPAUD_TABLE_PREFIX}_model_trainings",
344
+ ["id"],
345
+ unique=False,
346
+ )
347
+ op.create_table(
348
+ f"{LECRAPAUD_TABLE_PREFIX}_scores",
349
+ sa.Column("id", sa.BigInteger(), autoincrement=True, nullable=False),
350
+ sa.Column(
351
+ "created_at",
352
+ sa.TIMESTAMP(timezone=True),
353
+ server_default=sa.text("now()"),
354
+ nullable=False,
355
+ ),
356
+ sa.Column(
357
+ "updated_at",
358
+ sa.TIMESTAMP(timezone=True),
359
+ server_default=sa.text("now()"),
360
+ nullable=False,
361
+ ),
362
+ sa.Column("type", sa.String(length=50), nullable=False),
363
+ sa.Column("training_time", sa.Integer(), nullable=True),
364
+ sa.Column("eval_data_std", sa.Float(), nullable=True),
365
+ sa.Column("rmse", sa.Float(), nullable=True),
366
+ sa.Column("rmse_std_ratio", sa.Float(), nullable=True),
367
+ sa.Column("mae", sa.Float(), nullable=True),
368
+ sa.Column("mape", sa.Float(), nullable=True),
369
+ sa.Column("mam", sa.Float(), nullable=True),
370
+ sa.Column("mad", sa.Float(), nullable=True),
371
+ sa.Column("mae_mam_ratio", sa.Float(), nullable=True),
372
+ sa.Column("mae_mad_ratio", sa.Float(), nullable=True),
373
+ sa.Column("r2", sa.Float(), nullable=True),
374
+ sa.Column("logloss", sa.Float(), nullable=True),
375
+ sa.Column("accuracy", sa.Float(), nullable=True),
376
+ sa.Column("precision", sa.Float(), nullable=True),
377
+ sa.Column("recall", sa.Float(), nullable=True),
378
+ sa.Column("f1", sa.Float(), nullable=True),
379
+ sa.Column("roc_auc", sa.Float(), nullable=True),
380
+ sa.Column("threshold", sa.Float(), nullable=True),
381
+ sa.Column("precision_at_threshold", sa.Float(), nullable=True),
382
+ sa.Column("recall_at_threshold", sa.Float(), nullable=True),
383
+ sa.Column("model_training_id", sa.BigInteger(), nullable=False),
384
+ sa.ForeignKeyConstraint(
385
+ ["model_training_id"],
386
+ [f"{LECRAPAUD_TABLE_PREFIX}_model_trainings.id"],
387
+ ondelete="CASCADE",
388
+ ),
389
+ sa.PrimaryKeyConstraint("id"),
390
+ )
391
+ op.create_index(
392
+ op.f(f"ix_{LECRAPAUD_TABLE_PREFIX}_scores_id"),
393
+ f"{LECRAPAUD_TABLE_PREFIX}_scores",
394
+ ["id"],
395
+ unique=False,
396
+ )
189
397
  # ### end Alembic commands ###
190
398
 
191
399
 
192
400
  def downgrade() -> None:
193
401
  # ### commands auto generated by Alembic - please adjust! ###
194
- op.drop_index(op.f('ix_lecrapaud_scores_id'), table_name='lecrapaud_scores')
195
- op.drop_table('lecrapaud_scores')
196
- op.drop_index(op.f('ix_lecrapaud_model_trainings_id'), table_name='lecrapaud_model_trainings')
197
- op.drop_table('lecrapaud_model_trainings')
198
- op.drop_index(op.f('ix_lecrapaud_feature_selection_ranks_id'), table_name='lecrapaud_feature_selection_ranks')
199
- op.drop_table('lecrapaud_feature_selection_ranks')
200
- op.drop_table('lecrapaud_feature_selection_association')
201
- op.drop_index(op.f('ix_lecrapaud_model_selections_id'), table_name='lecrapaud_model_selections')
202
- op.drop_table('lecrapaud_model_selections')
203
- op.drop_index(op.f('ix_lecrapaud_feature_selections_id'), table_name='lecrapaud_feature_selections')
204
- op.drop_table('lecrapaud_feature_selections')
205
- op.drop_table('lecrapaud_experiment_target_association')
206
- op.drop_index(op.f('ix_lecrapaud_targets_id'), table_name='lecrapaud_targets')
207
- op.drop_table('lecrapaud_targets')
208
- op.drop_index(op.f('ix_lecrapaud_models_id'), table_name='lecrapaud_models')
209
- op.drop_table('lecrapaud_models')
210
- op.drop_index(op.f('ix_lecrapaud_features_id'), table_name='lecrapaud_features')
211
- op.drop_table('lecrapaud_features')
212
- op.drop_index(op.f('ix_lecrapaud_experiments_id'), table_name='lecrapaud_experiments')
213
- op.drop_table('lecrapaud_experiments')
402
+ op.drop_index(
403
+ op.f(f"ix_{LECRAPAUD_TABLE_PREFIX}_scores_id"),
404
+ table_name=f"{LECRAPAUD_TABLE_PREFIX}_scores",
405
+ )
406
+ op.drop_table(f"{LECRAPAUD_TABLE_PREFIX}_scores")
407
+ op.drop_index(
408
+ op.f(f"ix_{LECRAPAUD_TABLE_PREFIX}_model_trainings_id"),
409
+ table_name=f"{LECRAPAUD_TABLE_PREFIX}_model_trainings",
410
+ )
411
+ op.drop_table(f"{LECRAPAUD_TABLE_PREFIX}_model_trainings")
412
+ op.drop_index(
413
+ op.f(f"ix_{LECRAPAUD_TABLE_PREFIX}_feature_selection_ranks_id"),
414
+ table_name=f"{LECRAPAUD_TABLE_PREFIX}_feature_selection_ranks",
415
+ )
416
+ op.drop_table(f"{LECRAPAUD_TABLE_PREFIX}_feature_selection_ranks")
417
+ op.drop_table(f"{LECRAPAUD_TABLE_PREFIX}_feature_selection_association")
418
+ op.drop_index(
419
+ op.f(f"ix_{LECRAPAUD_TABLE_PREFIX}_model_selections_id"),
420
+ table_name=f"{LECRAPAUD_TABLE_PREFIX}_model_selections",
421
+ )
422
+ op.drop_table(f"{LECRAPAUD_TABLE_PREFIX}_model_selections")
423
+ op.drop_index(
424
+ op.f(f"ix_{LECRAPAUD_TABLE_PREFIX}_feature_selections_id"),
425
+ table_name=f"{LECRAPAUD_TABLE_PREFIX}_feature_selections",
426
+ )
427
+ op.drop_table(f"{LECRAPAUD_TABLE_PREFIX}_feature_selections")
428
+ op.drop_table(f"{LECRAPAUD_TABLE_PREFIX}_experiment_target_association")
429
+ op.drop_index(
430
+ op.f(f"ix_{LECRAPAUD_TABLE_PREFIX}_targets_id"),
431
+ table_name=f"{LECRAPAUD_TABLE_PREFIX}_targets",
432
+ )
433
+ op.drop_table(f"{LECRAPAUD_TABLE_PREFIX}_targets")
434
+ op.drop_index(
435
+ op.f(f"ix_{LECRAPAUD_TABLE_PREFIX}_models_id"),
436
+ table_name=f"{LECRAPAUD_TABLE_PREFIX}_models",
437
+ )
438
+ op.drop_table(f"{LECRAPAUD_TABLE_PREFIX}_models")
439
+ op.drop_index(
440
+ op.f(f"ix_{LECRAPAUD_TABLE_PREFIX}_features_id"),
441
+ table_name=f"{LECRAPAUD_TABLE_PREFIX}_features",
442
+ )
443
+ op.drop_table(f"{LECRAPAUD_TABLE_PREFIX}_features")
444
+ op.drop_index(
445
+ op.f(f"ix_{LECRAPAUD_TABLE_PREFIX}_experiments_id"),
446
+ table_name=f"{LECRAPAUD_TABLE_PREFIX}_experiments",
447
+ )
448
+ op.drop_table(f"{LECRAPAUD_TABLE_PREFIX}_experiments")
214
449
  # ### end Alembic commands ###
lecrapaud/db/alembic/versions/2025_06_24_1216-c62251b129ed_.py CHANGED
@@ -5,26 +5,31 @@ Revises: f089dfb7e3ba
5
5
  Create Date: 2025-06-24 12:16:21.949079
6
6
 
7
7
  """
8
+
8
9
  from typing import Sequence, Union
9
10
 
10
11
  from alembic import op
11
12
  import sqlalchemy as sa
13
+ from lecrapaud.config import LECRAPAUD_TABLE_PREFIX
12
14
 
13
15
 
14
16
  # revision identifiers, used by Alembic.
15
- revision: str = 'c62251b129ed'
16
- down_revision: Union[str, None] = 'f089dfb7e3ba'
17
+ revision: str = "c62251b129ed"
18
+ down_revision: Union[str, None] = "f089dfb7e3ba"
17
19
  branch_labels: Union[str, Sequence[str], None] = None
18
20
  depends_on: Union[str, Sequence[str], None] = None
19
21
 
20
22
 
21
23
  def upgrade() -> None:
22
24
  # ### commands auto generated by Alembic - please adjust! ###
23
- op.add_column('lecrapaud_experiments', sa.Column('context', sa.JSON(), nullable=True))
25
+ op.add_column(
26
+ f"{LECRAPAUD_TABLE_PREFIX}_experiments",
27
+ sa.Column("context", sa.JSON(), nullable=True),
28
+ )
24
29
  # ### end Alembic commands ###
25
30
 
26
31
 
27
32
  def downgrade() -> None:
28
33
  # ### commands auto generated by Alembic - please adjust! ###
29
- op.drop_column('lecrapaud_experiments', 'context')
34
+ op.drop_column(f"{LECRAPAUD_TABLE_PREFIX}_experiments", "context")
30
35
  # ### end Alembic commands ###
lecrapaud/db/alembic/versions/2025_06_24_1711-86457e2f333f_.py CHANGED
@@ -5,30 +5,41 @@ Revises: c62251b129ed
5
5
  Create Date: 2025-06-24 17:11:25.187876
6
6
 
7
7
  """
8
+
8
9
  from typing import Sequence, Union
9
10
 
10
11
  from alembic import op
11
12
  import sqlalchemy as sa
12
13
  from sqlalchemy.dialects import mysql
14
+ from lecrapaud.config import LECRAPAUD_TABLE_PREFIX
13
15
 
14
16
  # revision identifiers, used by Alembic.
15
- revision: str = '86457e2f333f'
16
- down_revision: Union[str, None] = 'c62251b129ed'
17
+ revision: str = "86457e2f333f"
18
+ down_revision: Union[str, None] = "c62251b129ed"
17
19
  branch_labels: Union[str, Sequence[str], None] = None
18
20
  depends_on: Union[str, Sequence[str], None] = None
19
21
 
20
22
 
21
23
  def upgrade() -> None:
22
24
  # ### commands auto generated by Alembic - please adjust! ###
23
- op.add_column('lecrapaud_scores', sa.Column('thresholds', sa.JSON(), nullable=True))
24
- op.add_column('lecrapaud_scores', sa.Column('f1_at_threshold', sa.Float(), nullable=True))
25
- op.drop_column('lecrapaud_scores', 'threshold')
25
+ op.add_column(
26
+ f"{LECRAPAUD_TABLE_PREFIX}_scores",
27
+ sa.Column("thresholds", sa.JSON(), nullable=True),
28
+ )
29
+ op.add_column(
30
+ f"{LECRAPAUD_TABLE_PREFIX}_scores",
31
+ sa.Column("f1_at_threshold", sa.Float(), nullable=True),
32
+ )
33
+ op.drop_column(f"{LECRAPAUD_TABLE_PREFIX}_scores", "threshold")
26
34
  # ### end Alembic commands ###
27
35
 
28
36
 
29
37
  def downgrade() -> None:
30
38
  # ### commands auto generated by Alembic - please adjust! ###
31
- op.add_column('lecrapaud_scores', sa.Column('threshold', mysql.FLOAT(), nullable=True))
32
- op.drop_column('lecrapaud_scores', 'f1_at_threshold')
33
- op.drop_column('lecrapaud_scores', 'thresholds')
39
+ op.add_column(
40
+ f"{LECRAPAUD_TABLE_PREFIX}_scores",
41
+ sa.Column("threshold", mysql.FLOAT(), nullable=True),
42
+ )
43
+ op.drop_column(f"{LECRAPAUD_TABLE_PREFIX}_scores", "f1_at_threshold")
44
+ op.drop_column(f"{LECRAPAUD_TABLE_PREFIX}_scores", "thresholds")
34
45
  # ### end Alembic commands ###
lecrapaud/db/alembic/versions/2025_06_25_1759-72aa496ca65b_.py CHANGED
@@ -5,26 +5,31 @@ Revises: 86457e2f333f
5
5
  Create Date: 2025-06-25 17:59:28.544283
6
6
 
7
7
  """
8
+
8
9
  from typing import Sequence, Union
9
10
 
10
11
  from alembic import op
11
12
  import sqlalchemy as sa
13
+ from lecrapaud.config import LECRAPAUD_TABLE_PREFIX
12
14
 
13
15
 
14
16
  # revision identifiers, used by Alembic.
15
- revision: str = '72aa496ca65b'
16
- down_revision: Union[str, None] = '86457e2f333f'
17
+ revision: str = "72aa496ca65b"
18
+ down_revision: Union[str, None] = "86457e2f333f"
17
19
  branch_labels: Union[str, Sequence[str], None] = None
18
20
  depends_on: Union[str, Sequence[str], None] = None
19
21
 
20
22
 
21
23
  def upgrade() -> None:
22
24
  # ### commands auto generated by Alembic - please adjust! ###
23
- op.add_column('lecrapaud_model_selections', sa.Column('best_thresholds', sa.JSON(), nullable=True))
25
+ op.add_column(
26
+ f"{LECRAPAUD_TABLE_PREFIX}_model_selections",
27
+ sa.Column("best_thresholds", sa.JSON(), nullable=True),
28
+ )
24
29
  # ### end Alembic commands ###
25
30
 
26
31
 
27
32
  def downgrade() -> None:
28
33
  # ### commands auto generated by Alembic - please adjust! ###
29
- op.drop_column('lecrapaud_model_selections', 'best_thresholds')
34
+ op.drop_column(f"{LECRAPAUD_TABLE_PREFIX}_model_selections", "best_thresholds")
30
35
  # ### end Alembic commands ###
lecrapaud/db/models/experiment.py CHANGED
@@ -18,24 +18,15 @@ from sqlalchemy import (
18
18
  from sqlalchemy.orm import relationship
19
19
 
20
20
  from lecrapaud.db.models.base import Base
21
- from lecrapaud.config import LECRAPAUD_TABLE_PREFIX
21
+ from lecrapaud.db.models.utils import create_association_table
22
22
 
23
23
  # jointures
24
- lecrapaud_experiment_target_association = Table(
25
- f"{LECRAPAUD_TABLE_PREFIX}_experiment_target_association",
26
- Base.metadata,
27
- Column(
28
- "experiment_id",
29
- BigInteger,
30
- ForeignKey(f"{LECRAPAUD_TABLE_PREFIX}_experiments.id", ondelete="CASCADE"),
31
- primary_key=True,
32
- ),
33
- Column(
34
- "target_id",
35
- BigInteger,
36
- ForeignKey(f"{LECRAPAUD_TABLE_PREFIX}_targets.id", ondelete="CASCADE"),
37
- primary_key=True,
38
- ),
24
+ lecrapaud_experiment_target_association = create_association_table(
25
+ name="experiment_target_association",
26
+ table1="experiments",
27
+ column1="experiment",
28
+ table2="targets",
29
+ column2="target"
39
30
  )
40
31
 
41
32
 
lecrapaud/db/models/feature_selection.py CHANGED
@@ -26,26 +26,15 @@ from collections.abc import Iterable
26
26
 
27
27
  from lecrapaud.db.session import get_db
28
28
  from lecrapaud.db.models.base import Base, with_db
29
- from lecrapaud.config import LECRAPAUD_TABLE_PREFIX
29
+ from lecrapaud.db.models.utils import create_association_table
30
30
 
31
31
  # jointures
32
- lecrapaud_feature_selection_association = Table(
33
- f"{LECRAPAUD_TABLE_PREFIX}_feature_selection_association",
34
- Base.metadata,
35
- Column(
36
- "feature_selection_id",
37
- BigInteger,
38
- ForeignKey(
39
- f"{LECRAPAUD_TABLE_PREFIX}_feature_selections.id", ondelete="CASCADE"
40
- ),
41
- primary_key=True,
42
- ),
43
- Column(
44
- "feature_id",
45
- BigInteger,
46
- ForeignKey(f"{LECRAPAUD_TABLE_PREFIX}_features.id", ondelete="CASCADE"),
47
- primary_key=True,
48
- ),
32
+ lecrapaud_feature_selection_association = create_association_table(
33
+ name="feature_selection_association",
34
+ table1="feature_selections",
35
+ column1="feature_selection",
36
+ table2="features",
37
+ column2="feature"
49
38
  )
50
39
 
51
40
 
lecrapaud/db/models/utils.py ADDED
@@ -0,0 +1,32 @@
1
+ from sqlalchemy import Table, Column, BigInteger, ForeignKey
2
+
3
+ from lecrapaud.db.models.base import Base
4
+ from lecrapaud.config import LECRAPAUD_TABLE_PREFIX
5
+
6
+
7
+ def create_association_table(name, table1, column1, table2, column2):
8
+ """Create an association table with the LECRAPAUD_TABLE_PREFIX.
9
+
10
+ Args:
11
+ name: The base name for the association table (will be prefixed)
12
+ table1: First table name (without prefix)
13
+ column1: First column name (will be used as f"{column1}_id")
14
+ table2: Second table name (without prefix)
15
+ column2: Second column name (will be used as f"{column2}_id")
16
+ """
17
+ return Table(
18
+ f"{LECRAPAUD_TABLE_PREFIX}_{name}",
19
+ Base.metadata,
20
+ Column(
21
+ f"{column1}_id",
22
+ BigInteger,
23
+ ForeignKey(f"{LECRAPAUD_TABLE_PREFIX}_{table1}.id", ondelete="CASCADE"),
24
+ primary_key=True,
25
+ ),
26
+ Column(
27
+ f"{column2}_id",
28
+ BigInteger,
29
+ ForeignKey(f"{LECRAPAUD_TABLE_PREFIX}_{table2}.id", ondelete="CASCADE"),
30
+ primary_key=True,
31
+ ),
32
+ )
lecrapaud/feature_engineering.py CHANGED
@@ -380,13 +380,16 @@ class PreprocessFeature:
380
380
  summary = summarize_dataframe(train)
381
381
  summary.to_csv(f"{self.experiment_dir}/feature_summary.csv", index=False)
382
382
 
383
+ # Save all features before selection
384
+ joblib.dump(
385
+ train, f"{self.preprocessing_dir}/all_features_before_selection.pkl"
386
+ )
387
+
383
388
  return train, val, test
384
389
 
385
390
  def inference(self):
386
391
  data = self.data
387
392
 
388
- print(data.shape)
389
- print(data.columns)
390
393
  # PCA
391
394
  if os.path.exists(f"{self.preprocessing_dir}/pcas.pkl"):
392
395
  pcas = joblib.load(f"{self.preprocessing_dir}/pcas.pkl")
@@ -655,7 +658,7 @@ class PreprocessFeature:
655
658
  X_transformed = pd.DataFrame(transformed, columns=column_names, index=df.index)
656
659
 
657
660
  # Try to convert columns to best possible dtypes
658
- X_transformed = X_transformed.convert_dtypes()
661
+ X_transformed = X_transformed.convert_ups()
659
662
 
660
663
  # Insert features in db
661
664
  if save_in_db:
lecrapaud/model_selection.py CHANGED
@@ -570,11 +570,13 @@ class ModelEngine:
570
570
  self.target_type = self._model.target_type
571
571
 
572
572
  # Load threshold
573
- self.threshold = (
574
- joblib.load(f"{self.path}/thresholds.pkl")
575
- if self.target_type == "classification"
576
- else None
577
- )
573
+ if (
574
+ os.path.exists(f"{self.path}/thresholds.pkl")
575
+ and self.target_type == "classification"
576
+ ):
577
+ self.threshold = joblib.load(f"{self.path}/thresholds.pkl")
578
+ else:
579
+ self.threshold = None
578
580
 
579
581
  logger.info(
580
582
  f"Loaded model {self._model.model_name} and threshold {self.threshold}"
{lecrapaud-0.14.5.dist-info → lecrapaud-0.14.6.dist-info}/METADATA RENAMED
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.3
2
2
  Name: lecrapaud
3
- Version: 0.14.5
3
+ Version: 0.14.6
4
4
  Summary: Framework for machine and deep learning, with regression, classification and time series analysis
5
5
  License: Apache License
6
6
  Author: Pierre H. Gallet
{lecrapaud-0.14.5.dist-info → lecrapaud-0.14.6.dist-info}/RECORD RENAMED
@@ -5,26 +5,27 @@ lecrapaud/db/__init__.py,sha256=82o9fMfaqKXPh2_rt44EzNRVZV1R4LScEnQYvj_TjK0,34
5
5
  lecrapaud/db/alembic/README,sha256=MVlc9TYmr57RbhXET6QxgyCcwWP7w-vLkEsirENqiIQ,38
6
6
  lecrapaud/db/alembic/env.py,sha256=RvTTBa3bDVBxmDtapAfzUoeWBgmVQU3s9U6HmQCAP84,2421
7
7
  lecrapaud/db/alembic/script.py.mako,sha256=MEqL-2qATlST9TAOeYgscMn1uy6HUS9NFvDgl93dMj8,635
8
- lecrapaud/db/alembic/versions/2025_06_23_1748-f089dfb7e3ba_.py,sha256=MNPyqWaQSHNV8zljD1G9f-LzrVz-nOKlgOhHEE0U8Oo,13060
9
- lecrapaud/db/alembic/versions/2025_06_24_1216-c62251b129ed_.py,sha256=g6aLRV6jAKXkPUEcs9FAeGfsYpe9rMTxfqbNib3U0-U,809
10
- lecrapaud/db/alembic/versions/2025_06_24_1711-86457e2f333f_.py,sha256=dl6tfvcqErgJ6NKvjve0euu7l0BWyEAKSS-ychsEAl8,1139
11
- lecrapaud/db/alembic/versions/2025_06_25_1759-72aa496ca65b_.py,sha256=sBgPLvvqI_HmPqQ0Kime1ZL1AHSeuYJHlmFJOnXWeuU,835
8
+ lecrapaud/db/alembic/versions/2025_06_23_1748-f089dfb7e3ba_.py,sha256=hyPW0Mt_B4ZAHnJYLREy7MAncNDLnEIyJQJW2pyz_LY,17228
9
+ lecrapaud/db/alembic/versions/2025_06_24_1216-c62251b129ed_.py,sha256=6Pf36HAXEVrVlnrohAe2O7gVaXpDiv3LLIP_EEgTyA0,917
10
+ lecrapaud/db/alembic/versions/2025_06_24_1711-86457e2f333f_.py,sha256=KjwjYvFaNqYmBLTYel8As37fyaBtNVWTqN_3M7y_2eI,1357
11
+ lecrapaud/db/alembic/versions/2025_06_25_1759-72aa496ca65b_.py,sha256=MiqooJuZ1etExl2he3MniaEv8G0LrmqY-0m22m9xKmc,943
12
12
  lecrapaud/db/alembic.ini,sha256=TXrZB4pWVLn2EUg867yp6paA_19vGeirO95mTPA3nbs,3699
13
13
  lecrapaud/db/models/__init__.py,sha256=Lhyw9fVLdom0Fc6yIP-ip8FjkU1EwVwjae5q2VM815Q,740
14
14
  lecrapaud/db/models/base.py,sha256=J9ew-0z_-tnWAwhVvOmVDys2R6jPF_oSca_ny6wpXQE,7606
15
- lecrapaud/db/models/experiment.py,sha256=x7-zTuCCecE401f3wQ9l6u8HyGOKORGc0b4_z-FMriU,4219
15
+ lecrapaud/db/models/experiment.py,sha256=5vKttTRLCcb2oS9-BHZhGxRRckM8E3C8LBdW34tPJuQ,3964
16
16
  lecrapaud/db/models/feature.py,sha256=5o77O2FyRObnLOCGNj8kaPSGM3pLv1Ov6mXXHYkmnYY,1136
17
- lecrapaud/db/models/feature_selection.py,sha256=epvOkU3eyhSXvEpRePd9Ylxl3feDvMUf_v6-jeU7fRE,3469
17
+ lecrapaud/db/models/feature_selection.py,sha256=6z8wOmiBETI2Vc4fwWnEeTpMzkOISu0zDGAU_09kJc8,3192
18
18
  lecrapaud/db/models/feature_selection_rank.py,sha256=Ydsb_rAT58FoSH13wkGjGPByzsjPx3DITXgJ2jgZmow,2198
19
19
  lecrapaud/db/models/model.py,sha256=F0hyMjd4FFHCv6_arIWBEmBCGOfG3b6_uzU8ExtFE90,952
20
20
  lecrapaud/db/models/model_selection.py,sha256=zNCumJvhGLGmjA14_agLQ-ZFgXc_uoXhtmBnxUk5iM8,1971
21
21
  lecrapaud/db/models/model_training.py,sha256=jAIYPdwBln2jf593soLQ730uYrTfNK8zdG8TesOqmh0,1698
22
22
  lecrapaud/db/models/score.py,sha256=fSfXLt6Dm-8Fy9ku0urMT5Fa6zNqn4YqVnEO4o3zKVI,1669
23
23
  lecrapaud/db/models/target.py,sha256=DKnfeaLU8eT8J_oh_vuFo5-o1CaoXR13xBbswme6Bgk,1649
24
+ lecrapaud/db/models/utils.py,sha256=-a-nWWmpJ2XzidIxo2COVUTrGZIPYCfBzjhcszJj_bM,1109
24
25
  lecrapaud/db/session.py,sha256=K9dTyXmO-aF_2r9RRHsDsbW9_zLNDaOcchXlpiv7cSo,2719
25
26
  lecrapaud/directories.py,sha256=0LrANuDgbuneSLker60c6q2hmGnQ3mKHIztTGzTx6Gw,826
26
27
  lecrapaud/experiment.py,sha256=u1whBtRFEhjM2OtKwHuUAp6T_7JXZhf_90luUiV4nD8,2514
27
- lecrapaud/feature_engineering.py,sha256=zG7EM428fEIsSwdtqhp08ajNjrL9qmCqDgiYlMjTTj0,32476
28
+ lecrapaud/feature_engineering.py,sha256=sVVW2eSyEa3ll-Ftg7LME5a9z7fepDnHpSfZVQR6hdw,32577
28
29
  lecrapaud/feature_selection.py,sha256=6ry-oVPQHbipm1XSE5YsH7AY0lQFt4CFbWiHiRs1nxg,43593
29
30
  lecrapaud/integrations/openai_integration.py,sha256=hHLF3fk5Bps8KNbNrEL3NUFa945jwClE6LrLpuMZOd4,7459
30
31
  lecrapaud/jobs/__init__.py,sha256=ZkrsyTOR21c_wN7RY8jPhm8jCrL1oCEtTsf3VFIlQiE,292
@@ -35,10 +36,10 @@ lecrapaud/misc/tabpfn_tests.ipynb,sha256=VkgsCUJ30d8jaL2VaWtQAgb8ngHPNtPgnXLs7QQ
35
36
  lecrapaud/misc/test-gpu-bilstm.ipynb,sha256=4nLuZRJVe2kn6kEmauhRiz5wkWT9AVrYhI9CEk_dYUY,9608
36
37
  lecrapaud/misc/test-gpu-resnet.ipynb,sha256=27Vu7nYwujYeh3fOxBNCnKJn3MXNPKZU-U8oDDUbymg,4944
37
38
  lecrapaud/misc/test-gpu-transformers.ipynb,sha256=k6MBSs_Um1h4PykvE-LTBcdpbWLbIFST_xl_AFW2jgI,8444
38
- lecrapaud/model_selection.py,sha256=5Q7KJ-oKDcW5ILlFzk4MjHyywB8HlNM6txCFNofku18,67796
39
+ lecrapaud/model_selection.py,sha256=vV2oO_fhhJ4dozOQCn_ySJzQT5gMVtQbLR4Wf59bi3A,67885
39
40
  lecrapaud/search_space.py,sha256=-JkzuMhaomdwiWi4HvVQY5hiw3-oREemJA16tbwEIp4,34854
40
41
  lecrapaud/utils.py,sha256=JdBB1NvbNIx4y0Una-kSZdo1_ZEocc5hwyYFIZKHmGg,8305
41
- lecrapaud-0.14.5.dist-info/LICENSE,sha256=MImCryu0AnqhJE_uAZD-PIDKXDKb8sT7v0i1NOYeHTM,11350
42
- lecrapaud-0.14.5.dist-info/METADATA,sha256=pJTwfJ_uMxEkAFPsNHf_94yXlCyb3QEApRkV-BDvwAI,11047
43
- lecrapaud-0.14.5.dist-info/WHEEL,sha256=b4K_helf-jlQoXBBETfwnf4B04YC67LOev0jo4fX5m8,88
44
- lecrapaud-0.14.5.dist-info/RECORD,,
42
+ lecrapaud-0.14.6.dist-info/LICENSE,sha256=MImCryu0AnqhJE_uAZD-PIDKXDKb8sT7v0i1NOYeHTM,11350
43
+ lecrapaud-0.14.6.dist-info/METADATA,sha256=-ejSnXniWYfJoMDl11DrEsSKc3WVcc3Vbiza5nVQuTg,11047
44
+ lecrapaud-0.14.6.dist-info/WHEEL,sha256=b4K_helf-jlQoXBBETfwnf4B04YC67LOev0jo4fX5m8,88
45
+ lecrapaud-0.14.6.dist-info/RECORD,,