[TMVA] Fix failing rbdt_xgboost test

  * avoid warnings with opened file that is not closed

  * don't assume the number of features is in the `_features_count`
    attribute (that one doesn't exist with xgboost 2.0)

  * support the `"reg:squarederror"` target, which is the default
    regression target in xgboost 2.0

Author: guitargeek

bindings/pyroot/pythonizations/python/ROOT/_pythonization/_tmva/_tree_inference.py

@@ -12,12 +12,13 @@
 import cppyy
 
 
-def SaveXGBoost(self, xgb_model, key_name, output_path, num_inputs=None, tmp_path="/tmp", threshold_dtype="float"):
+def SaveXGBoost(self, xgb_model, key_name, output_path, num_inputs, tmp_path="/tmp", threshold_dtype="float"):
     # Extract objective
     objective_map = {
         "multi:softprob": "softmax",  # Naming the objective softmax is more common today
         "binary:logistic": "logistic",
         "reg:linear": "identity",
+        "reg:squarederror": "identity",
     }
     model_objective = xgb_model.objective
     if not model_objective in objective_map:
@@ -48,7 +49,8 @@ def SaveXGBoost(self, xgb_model, key_name, output_path, num_inputs=None, tmp_pat
 
     import json
 
-    forest = json.load(open(tmp_path, "r"))
+    with open(tmp_path, "r") as json_file:
+        forest = json.load(json_file)
 
     # Determine whether the model has a bias paramter and write bias trees
     if hasattr(xgb_model, "base_score") and "reg:" in model_objective:
@@ -96,16 +98,6 @@ def fill_arrays(node, index, inputs_base, thresholds_base):
         for i in range(num_trees):
             outputs[i] = int(i % num_outputs)
 
-    # Determine number of input variables
-    if not num_inputs is None:
-        pass
-    elif hasattr(xgb_model, "_features_count"):
-        num_inputs = xgb_model._features_count
-    else:
-        raise Exception(
-            "Failed to get number of input variables from XGBoost model. Please provide the additional keyword argument 'num_inputs' to this function."
-        )
-
     # Store arrays in a ROOT file in a folder with the given key name
     # TODO: Write single values as simple integers and not vectors.
     f = cppyy.gbl.TFile(output_path, "RECREATE")

tmva/tmva/test/rbdt_xgboost.py

@@ -21,7 +21,7 @@ def _test_XGBBinary(backend, label):
     x, y = create_dataset(1000, 10, 2)
     xgb = xgboost.XGBClassifier(n_estimators=100, max_depth=3)
     xgb.fit(x, y)
-    ROOT.TMVA.Experimental.SaveXGBoost(xgb, "myModel", "testXGBBinary{}.root".format(label))
+    ROOT.TMVA.Experimental.SaveXGBoost(xgb, "myModel", "testXGBBinary{}.root".format(label), num_inputs=10)
     bdt = ROOT.TMVA.Experimental.RBDT[backend]("myModel", "testXGBBinary{}.root".format(label))
 
     y_xgb = xgb.predict_proba(x)[:, 1].squeeze()
@@ -51,7 +51,7 @@ def _test_XGBMulticlass(backend, label):
     x, y = create_dataset(1000, 10, 3)
     xgb = xgboost.XGBClassifier(n_estimators=100, max_depth=3)
     xgb.fit(x, y)
-    ROOT.TMVA.Experimental.SaveXGBoost(xgb, "myModel", "testXGBMulticlass{}.root".format(label))
+    ROOT.TMVA.Experimental.SaveXGBoost(xgb, "myModel", "testXGBMulticlass{}.root".format(label), num_inputs=10)
     bdt = ROOT.TMVA.Experimental.RBDT[backend]("myModel", "testXGBMulticlass{}.root".format(label))
 
     y_xgb = xgb.predict_proba(x)