adaptions to reflect the adapted nuisance estimation for the PLR IV-type score

Author: MalteKurz

doc/examples/double_ml_bonus_data.ipynb

@@ -98,7 +98,7 @@
     "        'max_features': 'sqrt',\n",
     "        'max_depth': 5}\n",
     "\n",
-    "dml_plr_rf.set_ml_nuisance_params('ml_g', 'tg', pars)\n",
+    "dml_plr_rf.set_ml_nuisance_params('ml_l', 'tg', pars)\n",
     "dml_plr_rf.set_ml_nuisance_params('ml_m', 'tg', pars)"
    ]
   },
@@ -153,7 +153,7 @@
     "                                'dml2')\n",
     "\n",
     "# set some hyperparameters for the learners\n",
-    "dml_plr_lasso.set_ml_nuisance_params('ml_g', 'tg', {'alpha': 0.0005})\n",
+    "dml_plr_lasso.set_ml_nuisance_params('ml_l', 'tg', {'alpha': 0.0005})\n",
     "dml_plr_lasso.set_ml_nuisance_params('ml_m', 'tg', {'alpha': 0.0026})"
    ]
   },
@@ -265,7 +265,7 @@
  ],
  "metadata": {
   "kernelspec": {
-   "display_name": "Python 3",
+   "display_name": "Python 3 (ipykernel)",
    "language": "python",
    "name": "python3"
   },
@@ -279,7 +279,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.8.6"
+   "version": "3.9.7"
   }
  },
  "nbformat": 4,

doc/guide/basics.rst

@@ -176,7 +176,7 @@ other half of observations indexed with :math:`i \in I`
 
     .. ipython:: python
 
-        def non_orth_score(y, d, g_hat, m_hat, smpls):
+        def non_orth_score(y, d, l_hat, g_hat, m_hat, smpls):
             u_hat = y - g_hat
             psi_a = -np.multiply(d, d)
             psi_b = np.multiply(d, u_hat)
@@ -222,7 +222,7 @@ other half of observations indexed with :math:`i \in I`
 
     .. jupyter-execute::
 
-        non_orth_score = function(y, d, g_hat, m_hat, smpls) {
+        non_orth_score = function(y, d, l_hat, g_hat, m_hat, smpls) {
          u_hat = y - g_hat
          psi_a = -1*d*d
          psi_b = d*u_hat

doc/guide/learners.rst

@@ -81,7 +81,7 @@ Without further specification of the hyperparameters, default values are used. T
         dml_plr_obj = dml.DoubleMLPLR(obj_dml_data,
                                       RandomForestRegressor(),
                                       RandomForestRegressor())
-        dml_plr_obj.set_ml_nuisance_params('ml_g', 'd', {'n_estimators': 10});
+        dml_plr_obj.set_ml_nuisance_params('ml_l', 'd', {'n_estimators': 10});
         print(dml_plr_obj.fit().summary)
 
 Setting treatment-variable-specific or fold-specific hyperparameters:
@@ -163,7 +163,7 @@ In this case the tuning should be done externally and the parameters can then be
         ml_g = Lasso()
         ml_m = Lasso()
         dml_plr_obj = dml.DoubleMLPLR(dml_data, ml_g, ml_m)
-        dml_plr_obj.set_ml_nuisance_params('ml_g', 'd', {'alpha': ml_g_tune.alpha_});
+        dml_plr_obj.set_ml_nuisance_params('ml_l', 'd', {'alpha': ml_g_tune.alpha_});
         dml_plr_obj.set_ml_nuisance_params('ml_m', 'd', {'alpha': ml_m_tune.alpha_});
         print(dml_plr_obj.params)
         print(dml_plr_obj.fit().summary)

doc/guide/scores.rst

@@ -280,7 +280,7 @@ can be obtained with
 
         import numpy as np
 
-        def non_orth_score(y, d, g_hat, m_hat, smpls):
+        def non_orth_score(y, d, l_hat, g_hat, m_hat, smpls):
             u_hat = y - g_hat
             psi_a = -np.multiply(d, d)
             psi_b = np.multiply(d, u_hat)
@@ -290,7 +290,7 @@ can be obtained with
 
     .. jupyter-execute::
 
-        non_orth_score = function(y, d, g_hat, m_hat, smpls) {
+        non_orth_score = function(y, d, l_hat, g_hat, m_hat, smpls) {
             u_hat = y - g_hat
             psi_a = -1*d*d
             psi_b = d*u_hat