def setUp(self):
input_data = np.zeros((12, 2))
target = np.zeros((12, 1))
for i in range(0, 12):
input_data[i][0] = 0.1 * i
input_data[i][1] = 1 - 0.1 * i
target[i][0] = np.math.floor(i / 3)
self.mytree = DecisionTreeRegressor().fit(input_data, target)
prune(self.mytree.tree_, 0)
target = data['median_house_value'] / 100000
d_tree = DecisionTreeRegressor(max_depth=2)
d_tree.fit(input_features, target)
# Objekt d_tree (gesamter Baum) wird in dem tree_array gespeichert
tree_array = [d_tree]
# .tree = tree object -> aufrufen der Tree-Klasse, capacity=sowas wie Anzahl der Knoten
num_nodes = d_tree.tree_.capacity
index = 0
alpha = 0
k = 1
# While-Schleife: solange num_nodes >1, mache:
while num_nodes > 1:
#deepcopy = eine Kopie, die unabhängig vom Original ist
tree_array.append(copy.deepcopy(tree_array[k - 1]))
min_node_idx, min_gk = determine_alpha(tree_array[k].tree_)
prune(tree_array[k].tree_, min_node_idx)
num_nodes = sum(1 * (tree_array[k].tree_.n_node_samples != 0))
k += 1
if False:
for k in range(0, len(tree_array)):
export_graphviz(tree_array[k], out_file='tree' + str(k) + '.dot')