Skip to content

Commit 6ad95f9

Browse files
MalteKurzMalteKurz
committed
update user guide and example gallery: renaming of learner ml_g to ml_l and additional IV-type score for PLIV
1 parent d340469 commit 6ad95f9

16 files changed

+361
-196
lines changed

‎doc/examples/R_double_ml_pension.ipynb‎

Lines changed: 72 additions & 6 deletions
Large diffs are not rendered by default.

‎doc/examples/R_double_ml_pipeline.ipynb‎

Lines changed: 23 additions & 23 deletions
Original file line numberDiff line numberDiff line change
@@ -113,9 +113,9 @@
113113
"lgr::get_logger(\"mlr3\")$set_threshold(\"warn\")\n",
114114
"\n",
115115
"learner_lasso = lrn(\"regr.cv_glmnet\", s=\"lambda.min\")\n",
116-
"ml_g_lasso = learner_lasso$clone()\n",
116+
"ml_l_lasso = learner_lasso$clone()\n",
117117
"ml_m_lasso = learner_lasso$clone()\n",
118-
"class(ml_g_lasso)"
118+
"class(ml_l_lasso)"
119119
]
120120
},
121121
{
@@ -125,7 +125,7 @@
125125
"metadata": {},
126126
"outputs": [],
127127
"source": [
128-
"# Random forest learner for nuisance part ml_g\n",
128+
"# Random forest learner for nuisance part ml_l\n",
129129
"learner_forest_regr = lrn(\"regr.ranger\",\n",
130130
" num.trees=500, mtry=floor(sqrt(dim_x)),\n",
131131
" max.depth=5, min.node.size=2)\n",
@@ -136,9 +136,9 @@
136136
" mtry=floor(sqrt(dim_x)),\n",
137137
" max.depth=5, min.node.size=2)\n",
138138
"\n",
139-
"ml_g_forest = learner_forest_regr$clone()\n",
139+
"ml_l_forest = learner_forest_regr$clone()\n",
140140
"ml_m_forest = learner_forest_classif$clone()\n",
141-
"class(ml_g_forest)"
141+
"class(ml_l_forest)"
142142
]
143143
},
144144
{
@@ -158,7 +158,7 @@
158158
"source": [
159159
"set.seed(123)\n",
160160
"obj_dml_plr_sim = DoubleMLPLR$new(dml_data_sim,\n",
161-
" ml_g=ml_g_lasso,\n",
161+
" ml_l=ml_l_lasso,\n",
162162
" ml_m=ml_m_lasso)\n",
163163
"obj_dml_plr_sim$fit()\n",
164164
"print(obj_dml_plr_sim)"
@@ -173,7 +173,7 @@
173173
"source": [
174174
"set.seed(123)\n",
175175
"obj_dml_plr_bonus = DoubleMLPLR$new(dml_data_bonus,\n",
176-
" ml_g=ml_g_forest,\n",
176+
" ml_l=ml_l_forest,\n",
177177
" ml_m=ml_m_forest)\n",
178178
"obj_dml_plr_bonus$fit()\n",
179179
"print(obj_dml_plr_bonus)"
@@ -205,11 +205,11 @@
205205
"# Lasso learner\n",
206206
"library(mlr3pipelines)\n",
207207
"pipe_lasso = po(lrn(\"regr.cv_glmnet\"), s = \"lambda.min\")\n",
208-
"ml_g_lasso_pipe = as_learner(pipe_lasso)\n",
208+
"ml_l_lasso_pipe = as_learner(pipe_lasso)\n",
209209
"ml_m_lasso_pipe = as_learner(pipe_lasso)\n",
210210
"\n",
211211
"# Class of the lasso learner\n",
212-
"class(ml_g_lasso_pipe)"
212+
"class(ml_l_lasso_pipe)"
213213
]
214214
},
215215
{
@@ -219,7 +219,7 @@
219219
"metadata": {},
220220
"outputs": [],
221221
"source": [
222-
"# Random forest learner for nuisance part ml_g\n",
222+
"# Random forest learner for nuisance part ml_l\n",
223223
"pipe_forest_regr = po(lrn(\"regr.ranger\"),\n",
224224
" num.trees=500, mtry=floor(sqrt(dim_x)),\n",
225225
" max.depth=5, min.node.size=2)\n",
@@ -230,11 +230,11 @@
230230
" mtry=floor(sqrt(dim_x)),\n",
231231
" max.depth=5, min.node.size=2)\n",
232232
"\n",
233-
"ml_g_forest_pipe = as_learner(pipe_forest_regr)\n",
233+
"ml_l_forest_pipe = as_learner(pipe_forest_regr)\n",
234234
"ml_m_forest_pipe = as_learner(pipe_forest_classif)\n",
235235
"\n",
236236
"# Class of the random forest learners\n",
237-
"class(ml_g_forest_pipe)\n",
237+
"class(ml_l_forest_pipe)\n",
238238
"class(ml_m_forest_pipe)"
239239
]
240240
},
@@ -255,7 +255,7 @@
255255
"source": [
256256
"set.seed(123)\n",
257257
"obj_dml_plr_sim_pipe = DoubleMLPLR$new(dml_data_sim,\n",
258-
" ml_g=ml_g_lasso_pipe,\n",
258+
" ml_l=ml_l_lasso_pipe,\n",
259259
" ml_m=ml_m_lasso_pipe)\n",
260260
"obj_dml_plr_sim_pipe$fit()\n",
261261
"print(obj_dml_plr_sim_pipe)"
@@ -270,7 +270,7 @@
270270
"source": [
271271
"set.seed(123)\n",
272272
"obj_dml_plr_bonus_pipe = DoubleMLPLR$new(dml_data_bonus,\n",
273-
" ml_g=ml_g_forest_pipe,\n",
273+
" ml_l=ml_l_forest_pipe,\n",
274274
" ml_m=ml_m_forest_pipe)\n",
275275
"obj_dml_plr_bonus_pipe$fit()\n",
276276
"print(obj_dml_plr_bonus_pipe)"
@@ -410,7 +410,7 @@
410410
"# Initiate new DoubleML object and estimate with graph learner\n",
411411
"set.seed(123)\n",
412412
"obj_dml_plr_sim_pipe_ensemble = DoubleMLPLR$new(dml_data_sim,\n",
413-
" ml_g = ensemble_pipe_regr,\n",
413+
" ml_l = ensemble_pipe_regr,\n",
414414
" ml_m = ensemble_pipe_regr)\n",
415415
"obj_dml_plr_sim_pipe_ensemble$fit()\n",
416416
"print(obj_dml_plr_sim_pipe_ensemble)"
@@ -425,7 +425,7 @@
425425
"source": [
426426
"set.seed(123)\n",
427427
"obj_dml_plr_bonus_pipe_ensemble = DoubleMLPLR$new(dml_data_bonus,\n",
428-
" ml_g = ensemble_pipe_regr,\n",
428+
" ml_l = ensemble_pipe_regr,\n",
429429
" ml_m = ensemble_pipe_classif)\n",
430430
"obj_dml_plr_bonus_pipe_ensemble$fit()\n",
431431
"print(obj_dml_plr_bonus_pipe_ensemble)"
@@ -495,7 +495,7 @@
495495
"\n",
496496
"set.seed(123)\n",
497497
"obj_dml_plr_bonus_pipe = DoubleMLPLR$new(dml_data_bonus,\n",
498-
" ml_g=ml_g_forest,\n",
498+
" ml_l=ml_l_forest,\n",
499499
" ml_m=stacklrn)\n",
500500
"obj_dml_plr_bonus_pipe$fit()\n",
501501
"print(obj_dml_plr_bonus_pipe)"
@@ -607,7 +607,7 @@
607607
"source": [
608608
"set.seed(123)\n",
609609
"obj_dml_plr_bonus_pipe2 = DoubleMLPLR$new(dml_data_bonus,\n",
610-
" ml_g=ml_g_lasso,\n",
610+
" ml_l=ml_l_lasso,\n",
611611
" ml_m=glrn)\n",
612612
"obj_dml_plr_bonus_pipe2$fit()\n",
613613
"print(obj_dml_plr_bonus_pipe2)"
@@ -640,7 +640,7 @@
640640
"source": [
641641
"set.seed(123)\n",
642642
"obj_dml_plr_bonus_pipe3 = DoubleMLPLR$new(dml_data_bonus,\n",
643-
" ml_g=ml_g_lasso,\n",
643+
" ml_l=ml_l_lasso,\n",
644644
" ml_m=glrn)\n",
645645
"obj_dml_plr_bonus_pipe3$fit()\n",
646646
"print(obj_dml_plr_bonus_pipe3)"
@@ -711,7 +711,7 @@
711711
"tune_settings = list(terminator = trm(\"evals\", n_evals = 10),\n",
712712
" algorithm = tnr(\"grid_search\", resolution = 10),\n",
713713
" rsmp_tune = rsmp(\"cv\", folds = 5),\n",
714-
" measure = list(\"ml_g\" = msr(\"regr.mse\"),\n",
714+
" measure = list(\"ml_l\" = msr(\"regr.mse\"),\n",
715715
" \"ml_m\" = msr(\"regr.mse\")))"
716716
]
717717
},
@@ -725,9 +725,9 @@
725725
"# Initiate new DoubleML object and execute tuning with graph learner\n",
726726
"set.seed(123)\n",
727727
"obj_dml_plr_sim_pipe_tune = DoubleMLPLR$new(dml_data_sim,\n",
728-
" ml_g=glrn_lasso,\n",
728+
" ml_l=glrn_lasso,\n",
729729
" ml_m=glrn_lasso)\n",
730-
"obj_dml_plr_sim_pipe_tune$tune(param_set = list(\"ml_g\" = par_grids,\n",
730+
"obj_dml_plr_sim_pipe_tune$tune(param_set = list(\"ml_l\" = par_grids,\n",
731731
" \"ml_m\" = par_grids),\n",
732732
" tune_settings=tune_settings)"
733733
]
@@ -786,7 +786,7 @@
786786
"mimetype": "text/x-r-source",
787787
"name": "R",
788788
"pygments_lexer": "r",
789-
"version": "3.6.1"
789+
"version": "4.0.4"
790790
}
791791
},
792792
"nbformat": 4,

‎doc/examples/double_ml_bonus_data.ipynb‎

Lines changed: 22 additions & 22 deletions
Original file line numberDiff line numberDiff line change
@@ -78,20 +78,20 @@
7878
"metadata": {},
7979
"outputs": [],
8080
"source": [
81-
"# Set machine learning methods for m & g\n",
82-
"ml_g = RandomForestRegressor()\n",
81+
"# Set machine learning methods for m & l\n",
82+
"ml_l = RandomForestRegressor()\n",
8383
"ml_m = RandomForestRegressor()\n",
8484
"n_folds = 2\n",
8585
"n_rep = 10\n",
8686
"\n",
8787
"np.random.seed(3141)\n",
8888
"dml_plr_rf = dml.DoubleMLPLR(dml_data,\n",
89-
" ml_g,\n",
89+
" ml_l,\n",
9090
" ml_m,\n",
91-
" n_folds,\n",
92-
" n_rep,\n",
93-
" 'partialling out',\n",
94-
" 'dml2')\n",
91+
" n_folds=n_folds,\n",
92+
" n_rep=n_rep,\n",
93+
" score='partialling out',\n",
94+
" dml_procedure='dml2')\n",
9595
"\n",
9696
"# set some hyperparameters for the learners\n",
9797
"pars = {'n_estimators': 500,\n",
@@ -137,20 +137,20 @@
137137
"metadata": {},
138138
"outputs": [],
139139
"source": [
140-
"# Set machine learning methods for m & g\n",
141-
"ml_g = Lasso()\n",
140+
"# Set machine learning methods for m & l\n",
141+
"ml_l = Lasso()\n",
142142
"ml_m = Lasso()\n",
143143
"n_folds = 2\n",
144144
"n_rep = 10\n",
145145
"\n",
146146
"np.random.seed(3141)\n",
147147
"dml_plr_lasso = dml.DoubleMLPLR(dml_data_lasso,\n",
148-
" ml_g,\n",
148+
" ml_l,\n",
149149
" ml_m,\n",
150-
" n_folds,\n",
151-
" n_rep,\n",
152-
" 'partialling out',\n",
153-
" 'dml2')\n",
150+
" n_folds=n_folds,\n",
151+
" n_rep=n_rep,\n",
152+
" score='partialling out',\n",
153+
" dml_procedure='dml2')\n",
154154
"\n",
155155
"# set some hyperparameters for the learners\n",
156156
"dml_plr_lasso.set_ml_nuisance_params('ml_l', 'tg', {'alpha': 0.0005})\n",
@@ -191,10 +191,10 @@
191191
"dml_irm_rf = dml.DoubleMLIRM(dml_data,\n",
192192
" ml_g,\n",
193193
" ml_m,\n",
194-
" n_folds,\n",
195-
" n_rep,\n",
196-
" 'ATE',\n",
197-
" 'dml2')\n",
194+
" n_folds=n_folds,\n",
195+
" n_rep=n_rep,\n",
196+
" score='ATE',\n",
197+
" dml_procedure='dml2')\n",
198198
"\n",
199199
"# set some hyperparameters for the learners\n",
200200
"pars = {'n_estimators': 500,\n",
@@ -241,10 +241,10 @@
241241
"dml_irm_lasso = dml.DoubleMLIRM(dml_data_lasso,\n",
242242
" ml_g,\n",
243243
" ml_m,\n",
244-
" n_folds,\n",
245-
" n_rep,\n",
246-
" 'ATE',\n",
247-
" 'dml2')\n",
244+
" n_folds=n_folds,\n",
245+
" n_rep=n_rep,\n",
246+
" score='ATE',\n",
247+
" dml_procedure='dml2')\n",
248248
"\n",
249249
"# set some hyperparameters for the learners\n",
250250
"dml_irm_lasso.set_ml_nuisance_params('ml_g0', 'tg', {'alpha': 0.0019})\n",

‎doc/examples/py_double_ml_multiway_cluster.ipynb‎

Lines changed: 6 additions & 6 deletions
Original file line numberDiff line numberDiff line change
@@ -190,15 +190,15 @@
190190
},
191191
"outputs": [],
192192
"source": [
193-
"# Set machine learning methods for m, g & r\n",
193+
"# Set machine learning methods for l, m & r\n",
194194
"learner = LassoCV()\n",
195-
"ml_g = clone(learner)\n",
195+
"ml_l = clone(learner)\n",
196196
"ml_m = clone(learner)\n",
197197
"ml_r = clone(learner)\n",
198198
"\n",
199199
"# initialize the DoubleMLPLIV object\n",
200200
"dml_pliv_obj = DoubleMLPLIV(obj_dml_data,\n",
201-
" ml_g, ml_m, ml_r,\n",
201+
" ml_l, ml_m, ml_r,\n",
202202
" n_folds=3)"
203203
]
204204
},
@@ -472,15 +472,15 @@
472472
},
473473
"outputs": [],
474474
"source": [
475-
"# Set machine learning methods for m & g\n",
475+
"# Set machine learning methods for l, m & r\n",
476476
"learner = LassoCV()\n",
477-
"ml_g = clone(learner)\n",
477+
"ml_l = clone(learner)\n",
478478
"ml_m = clone(learner)\n",
479479
"ml_r = clone(learner)\n",
480480
"\n",
481481
"# initialize the DoubleMLPLIV object\n",
482482
"dml_pliv_obj = DoubleMLPLIV(obj_dml_data,\n",
483-
" ml_g, ml_m, ml_r,\n",
483+
" ml_l, ml_m, ml_r,\n",
484484
" n_folds=3)"
485485
]
486486
},

0 commit comments

Comments
 (0)