size = int(len(test_data) / depth)
num_trees = min(depth, 11)
if num_trees % 2 == 0:
num_trees += 1
classes = list(set(data_point[0] for data_point in train_data))
classes.sort()
n_classes = len(classes)
confusion_matrix = [[0 for i in range(n_classes)] for j in range(n_classes)]
trees = []
num_samples = int(len(train_data) * 0.8)
data_samples = split_data(train_data, num_trees, num_samples)
for i in range(num_trees):
tree = DecisionTree.build_tree(data_samples[i], depth, size)
trees.append(tree)
for data_point in test_data:
predictions = []
for i in range(num_trees):
prediction = DecisionTree.predict(trees[i], data_point)
predictions.append(prediction)
prediction = max(set(predictions), key=predictions.count)
confusion_matrix[int(data_point[0]) - 1][int(prediction) - 1] += 1
DecisionTree.print_matrix(confusion_matrix)
DecisionTree.F_1(confusion_matrix)
