Implement copilot review suggestions

Author: maurever

h2o-algos/src/main/java/hex/glm/GLM.java

@@ -4109,11 +4109,16 @@ protected void updateProgress(boolean canScore) {
           _scoringHistory.addIterationScore(_state._iter, task._likelihood, objectiveControlVal);
         } else if (_model._parms._remove_offset_effects) {
             _scoringHistoryUnrestrictedModel.addIterationScore(_state._iter, _state.likelihood(), _state.objective());
-            Frame frameWithoutOffset = _state._dinfo._adaptedFrame.deepCopy("frWithoutOffsetForScoringOnly");
-            Vec offset = frameWithoutOffset.remove(_model._parms._offset_column);
-            GLMResDevTask task = new GLMResDevTask(_job._key,_dinfo,_parms, _model.beta()).doAll(frameWithoutOffset);
-            offset.remove();
-            frameWithoutOffset.remove();
+            Frame frameZeroOffset = _state._dinfo._adaptedFrame.deepCopy("frWithoutOffsetForScoringOnly");
+            int offsetIdx = frameZeroOffset.find(_model._parms._offset_column);
+            assert offsetIdx > -1;
+            Vec originalOffset = frameZeroOffset.vec(offsetIdx);
+            Vec zeroOffset = originalOffset.makeZero();
+            frameZeroOffset.replace(offsetIdx, zeroOffset).remove();
+            GLMResDevTask task = new GLMResDevTask(_job._key,_dinfo,_parms, _model.beta()).doAll(frameZeroOffset);
+            originalOffset.remove();
+            zeroOffset.remove();
+            frameZeroOffset.remove();
             double objectiveOffset = _state.objective(_model.beta(), task._likelihood);
             _scoringHistory.addIterationScore(_state._iter, task._likelihood, objectiveOffset); 
         } else {

h2o-docs/src/product/data-science/algo-params/control_variables.rst

@@ -21,9 +21,9 @@ To get the unrestricted model with its own metrics use ``glm.make_unrestricted_g
 
 The control variables' coefficients are set to zero in the variable importance table. Use the unrestricted model to get the variable importance table with all variables included. 
 
-If you set up the ``control_variables`` together with the ``remove_offset_effects`` feature, model metrics and scoring history are calculated with both effects enabled. 
-If you need to get a model with only one feature enabled, you can get it using ``glm.make_derived_glm_model(remove_control_variables_effects=True)`` or ``glm.make_derived_glm_model(remove_offset_effects=True)``
-If both features are enabled and ``score_each_iteration=True`` or ``generate_scoring_history=True``, the training the model with big data can be slowed down. The complexity is four times higher than the standard GLM metric calculation.
+If you set up the ``control_variables`` together with the ``remove_offset_effects`` feature, model metrics and scoring history are calculated with both features enabled (that is, with both offset and control-variable effects removed during scoring).  
+If you need to get a model with only one feature enabled, you can get it using ``glm.make_derived_glm_model(remove_control_variables_effects=True)`` or ``glm.make_derived_glm_model(remove_offset_effects=True)``.
+If both features are enabled and ``score_each_iteration=True`` or ``generate_scoring_history=True``, training the model on big data can be slowed down. The complexity is four times higher than the standard GLM metric calculation.
 
 **Notes**:
 

h2o-docs/src/product/data-science/algo-params/remove_offset_effects.rst

@@ -13,9 +13,9 @@ Model metrics and scoring history are calculated for both the restricted model (
 
 To get the unrestricted model with its own metrics use ``glm.make_unrestricted_glm_model()`` / ``h2o.make_unrestricted_glm_model(glm)``.
 
-If you set up the ``remove_offset_effects`` together with the ``control_variables`` feature, model metrics and scoring history are calculated with both effects enabled.  
-If you need to get a model with only one feature enabled, you can get it using ``glm.make_derived_glm_model(remove_control_variables_effects=True)`` or ``glm.make_derived_glm_model(remove_offset_effects=True)``
-If both features are enabled and ``score_each_iteration=True`` or ``generate_scoring_history=True``, the training the model with big data can be slowed down. The complexity is four times higher than the standard GLM metric calculation.
+If you set up the ``remove_offset_effects`` together with the ``control_variables`` model metrics and scoring history are calculated with both features enabled (that is, with both offset and control-variable effects removed during scoring).  
+If you need to get a model with only one feature enabled, you can get it using ``glm.make_derived_glm_model(remove_control_variables_effects=True)`` or ``glm.make_derived_glm_model(remove_offset_effects=True)``.
+If both features are enabled and ``score_each_iteration=True`` or ``generate_scoring_history=True``, training the model on big data can be slowed down. The complexity is four times higher than the standard GLM metric calculation.
 
 **Notes**:
 

h2o-py/tests/testdir_algos/glm/pyunit_glm_control_variables_grid_search.py

@@ -24,7 +24,7 @@ def glm_grid_search_control_variables():
     train['wt_100'] = (train['CAPSULE'] == "1").ifelse(100, 1)
 
     hyper_parameters = OrderedDict()
-    hyper_parameters["control_variables"] = [["wt_2"], [["wt_100"]]]
+    hyper_parameters["control_variables"] = [["wt_2"], ["wt_100"]]
     print("GLM grid with the following hyper_parameters:", hyper_parameters)
 
     gs = H2OGridSearch(H2OGeneralizedLinearEstimator, hyper_params=hyper_parameters)