def classify_data_item(self, data_item, decision_tree):
"""
Function that classifies a data item from test data set by traversing the trained decision tree.
:param data_item:
:param decision_tree:
:return:
"""
test_data = TestData()
current_node = decision_tree
split_feature_value = data_item[current_node.split_feature_index]
# Traverse tree until it finds the leaf node or the split on the feature value is not found.
while len(current_node.children
) > 0 and split_feature_value in current_node.children:
current_node = current_node.children[split_feature_value]
split_feature_value = data_item[current_node.split_feature_index]
classified = current_node.class_label
test_data.predicted_class = classified
test_data.is_error = True if data_item[0] != classified else False
test_data.is_false_negative = True if data_item[0] == GlobalVectors.POSITIVE_VALUE \
and classified == GlobalVectors.NEGATIVE_VALUE else False
test_data.is_true_negative = True if data_item[0] == GlobalVectors.NEGATIVE_VALUE \
and classified == GlobalVectors.NEGATIVE_VALUE else False
test_data.is_false_positive = True if data_item[0] == GlobalVectors.NEGATIVE_VALUE \
and classified == GlobalVectors.POSITIVE_VALUE else False
test_data.is_true_positive = True if data_item[0] == GlobalVectors.POSITIVE_VALUE \
and classified == GlobalVectors.POSITIVE_VALUE else False
return test_data
def classify_data_item(self, data_item, learn_tree_list):
"""
Function that classifies a data item from test data set by traversing the trained decision tree.
:param data_item:
:param decision_tree:
:return:
"""
result_dict = dict()
test_data = TestData()
# Iterating to all the learnt bag trees to find majority vote.
for ltree in learn_tree_list:
current_node = ltree.decision_tree
split_feature_value = data_item[current_node.split_feature_index]
# Traverse tree until it finds the leaf node or the split on the feature value is not found.
while len(current_node.children
) > 0 and split_feature_value in current_node.children:
current_node = current_node.children[split_feature_value]
split_feature_value = data_item[
current_node.split_feature_index]
if current_node.class_label not in result_dict:
result_dict[current_node.class_label] = 1
else:
result_dict[current_node.class_label] += 1
# Getting the majority vote here.
classified = self.get_majority_vote(result_dict)
test_data.predicted_class = classified
test_data.is_error = True if data_item[0] != classified else False
test_data.is_false_negative = True if data_item[0] == GlobalVectors.POSITIVE_VALUE \
and classified == GlobalVectors.NEGATIVE_VALUE else False
test_data.is_true_negative = True if data_item[0] == GlobalVectors.NEGATIVE_VALUE \
and classified == GlobalVectors.NEGATIVE_VALUE else False
test_data.is_false_positive = True if data_item[0] == GlobalVectors.NEGATIVE_VALUE \
and classified == GlobalVectors.POSITIVE_VALUE else False
test_data.is_true_positive = True if data_item[0] == GlobalVectors.POSITIVE_VALUE \
and classified == GlobalVectors.POSITIVE_VALUE else False
return test_data