Improve regular logistic regression

Author: pobonomo

docs/examples/example2_student_admission.py

@@ -123,7 +123,7 @@
 
 # Run our regression
 scaler = StandardScaler()
-regression = LogisticRegression(random_state=1)
+regression = LogisticRegression(random_state=10)
 pipe = make_pipeline(scaler, regression)
 pipe.fit(X=historical_data.loc[:, features], y=historical_data.loc[:, target])
 
@@ -144,7 +144,7 @@
 nstudents = 20
 
 # Select randomly nstudents in the data
-studentsdata = studentsdata.sample(nstudents, random_state=1)
+studentsdata = studentsdata.sample(nstudents, random_state=10)
 
 
 ######################################################################
@@ -229,7 +229,8 @@
 # documentation <https://gurobi-machinelearning.readthedocs.io/en/v1.3.0/mlm-examples/student_admission.html>`__
 # for dealing with those approximations.
 #
-
+m.Params.NodeLimit = 10000
+m.write("students.lp")
 m.optimize()
 
 
@@ -251,7 +252,7 @@
 # regression in a solution as a pandas dataframe using input_values.
 #
 
-pred_constr.input_values
+pred_constr.input_values.round(3)
 
 
 ######################################################################

src/gurobi_ml/modeling/softmax.py

@@ -100,17 +100,22 @@ def _addGenConstrIndicatorMvarV10(binvar, binval, lhs, sense, rhs, name):
 def logistic(predictor_model: AbstractPredictorConstr, linear_predictor: gp.MVar):
     log_result = predictor_model.output[:, 1]
 
-    for index in np.ndindex(log_result.shape):
-        predictor_model.gp_model.addGenConstrLogistic(
-            linear_predictor[index],
-            log_result[index],
-            name=predictor_model._indexed_name(index, "logistic"),
+    if _HAS_NL_EXPR:
+        predictor_model.gp_model.addConstr(
+            log_result == nlfunc.logistic(linear_predictor[:, 0])
         )
-    num_gc = predictor_model.gp_model.NumGenConstrs
-    predictor_model.gp_model.update()
-    for gen_constr in predictor_model.gp_model.getGenConstrs()[num_gc:]:
-        for attr, val in predictor_model.attributes.items():
-            gen_constr.setAttr(attr, val)
+    else:
+        for index in np.ndindex(log_result.shape):
+            predictor_model.gp_model.addGenConstrLogistic(
+                linear_predictor[index],
+                log_result[index],
+                name=predictor_model._indexed_name(index, "logistic"),
+            )
+        num_gc = predictor_model.gp_model.NumGenConstrs
+        predictor_model.gp_model.update()
+        for gen_constr in predictor_model.gp_model.getGenConstrs()[num_gc:]:
+            for attr, val in predictor_model.attributes.items():
+                gen_constr.setAttr(attr, val)
 
 
 def hardmax(