def test_random_forest_classifier_mismatched_estimator_counts(self):
model = RandomForestClassifier(n_estimators=3)
X = [[0, 1], [1, 1], [2, 0]]
X = numpy.array(X, dtype=numpy.float32)
y = ['A', 'B', 'A']
model.fit(X, y)
# Training code can manipulate n_estimators causing
# n_estimators != len(estimators_). So simulate that here.
model.n_estimators += 1
model_onnx, prefix = convert_model(model, 'binary classifier',
[('input',
FloatTensorType([None, 2]))])
dump_data_and_model(X, model, model_onnx,
basename=prefix + "Bin" +
model.__class__.__name__ +
'_mismatched_estimator_counts')
def test_random_forest_regressor_mismatches(self):
iris = load_iris()
X, y = iris.data, iris.target
X_train, X_test, y_train, _ = train_test_split(X, y, random_state=13)
X_test = X_test.astype(numpy.float32)
clr = RandomForestRegressor(n_jobs=1, n_estimators=100)
clr.fit(X_train, y_train)
clr.fit(X, y)
model_onnx, prefix = convert_model(
clr, 'reg', [('input', FloatTensorType([None, 4]))])
dump_data_and_model(X_test,
clr,
model_onnx,
basename=prefix + "RegMis" +
clr.__class__.__name__ +
'_mismatched_estimator_counts')
def test_random_forest_regressor_mismatched_estimator_counts(self):
model = RandomForestRegressor(n_estimators=3)
X = [[0, 1], [1, 1], [2, 0]]
X = numpy.array(X, dtype=numpy.float32)
y = numpy.array([100, -10, 50], dtype=numpy.float32)
model.fit(X, y)
# Training code can manipulate n_estimators causing
# n_estimators != len(estimators_). So simulate that here.
model.n_estimators += 1
model_onnx, prefix = convert_model(model, 'single regressor',
[('input',
FloatTensorType([None, 2]))])
dump_data_and_model(X, model, model_onnx,
basename=prefix + "Reg" +
model.__class__.__name__ +
"_mismatched_estimator_counts")
