def ada_classify(tran_data, test_data):
'''
使用训练的adaboost决策树对测试数据进行预测
'''
res_cls = []
sub_tree_wh = []
wh_classifier = adaboost(tran_data, 300)
final_cls = []
'''
[ [权重，决策树],
...
]
'''
for wh_tree in wh_classifier:
sub_tree_wh.append(wh_tree[0])
res_cls.append(wh_tree[1].classify(test_data))
clses_T = map(list, zip(*res_cls))
print "样本长度", len(clses_T), "分类器个数", len(res_cls)
for c in clses_T:
vote_res = {}
for i, wh in zip(c, sub_tree_wh):
if i in vote_res:
vote_res[i] += wh
else:
vote_res[i] = wh
final_cls.append(max(vote_res, key=vote_res.get))
print "分类结果:", final_cls
accurcy = check_accurcy(test_data, final_cls)
return accurcy
def ada_classify(tran_data, test_data):
'''
使用训练的adaboost决策树对测试数据进行预测
'''
res_cls = []
sub_tree_wh = []
wh_classifier = adaboost(tran_data,300)
final_cls = []
'''
[ [权重，决策树],
...
]
'''
for wh_tree in wh_classifier:
sub_tree_wh.append(wh_tree[0])
res_cls.append(wh_tree[1].classify(test_data))
clses_T = map(list, zip(*res_cls))
print "样本长度", len(clses_T), "分类器个数", len(res_cls)
for c in clses_T:
vote_res = {}
for i, wh in zip(c, sub_tree_wh):
if i in vote_res:
vote_res[i] += wh
else:
vote_res[i] = wh
final_cls.append(max(vote_res, key=vote_res.get))
print "分类结果:", final_cls
accurcy = check_accurcy(test_data, final_cls)
return accurcy
def rd_fr_classify(tran_data, test_data):
forests = random_fr(tran_data)
res_clses = []
cls = []
for tree in forests:
res_clses.append(tree.classify(test_data))
clses_T = map(list, zip(*res_clses))
for c in clses_T:
vote_cls = collections.Counter(c).most_common(1)[0][0]
cls.append(vote_cls)
accurcy = check_accurcy(test_data, cls)
return accurcy
def rd_fr_classify(tran_data, test_data):
forests = random_fr(tran_data)
res_clses = []
cls = []
for tree in forests:
res_clses.append(tree.classify(test_data))
clses_T = map(list, zip(*res_clses))
for c in clses_T:
vote_cls = collections.Counter(c).most_common(1)[0][0]
cls.append(vote_cls)
accurcy = check_accurcy(test_data, cls)
return accurcy
self.leaf_err_sum(cur_node, leaf_err_set)
leaf_e_sum = sum(leaf_err_set) + 0.5 * len(leaf_err_set)
leaf_err_ratio = leaf_e_sum / len(cur_node.dataset)
std_dev = np.sqrt(leaf_err_ratio * (1 - leaf_err_ratio))
if leaf_e_sum + std_dev > cur_err_sum:
print leaf_e_sum + std_dev, cur_err_sum, " prun!!!!"
cur_node.childNode = {}
cur_node.cls = get_cls_from_data(cur_node.dataset)
else:
for _, c in cur_node.childNode.items():
self.__prun_tree(c)
def prun_tree(self):
self.__prun_tree(self.root)
if __name__ == '__main__':
#dataset = read_data("test.txt")
#dataset = read_data("breast-cancer-assignment5.txt")
dataset = read_data("german-assignment5.txt")
attr_set = range(len(dataset[0]))
DiscType = get_disc_val(dataset)
decisin_tree = DecisionTree(dataset[1:],attr_set, DiscType)
#decisin_tree.prun_tree()
res_cls = decisin_tree.classify(dataset[1:])
#res_cls = decisin_tree.classify(dataset[1:])
#print res_cls
acc = check_accurcy(dataset[1:], res_cls)
print acc
self.leaf_err_sum(cur_node, leaf_err_set)
leaf_e_sum = sum(leaf_err_set) + 0.5 * len(leaf_err_set)
leaf_err_ratio = leaf_e_sum / len(cur_node.dataset)
std_dev = np.sqrt(leaf_err_ratio * (1 - leaf_err_ratio))
if leaf_e_sum + std_dev > cur_err_sum:
print leaf_e_sum + std_dev, cur_err_sum, " prun!!!!"
cur_node.childNode = {}
cur_node.cls = get_cls_from_data(cur_node.dataset)
else:
for _, c in cur_node.childNode.items():
self.__prun_tree(c)
def prun_tree(self):
self.__prun_tree(self.root)
if __name__ == '__main__':
#dataset = read_data("test.txt")
#dataset = read_data("breast-cancer-assignment5.txt")
dataset = read_data("german-assignment5.txt")
attr_set = range(len(dataset[0]))
DiscType = get_disc_val(dataset)
decisin_tree = DecisionTree(dataset[1:], attr_set, DiscType)
#decisin_tree.prun_tree()
res_cls = decisin_tree.classify(dataset[1:])
#res_cls = decisin_tree.classify(dataset[1:])
#print res_cls
acc = check_accurcy(dataset[1:], res_cls)
print acc
