#! /usr/bin/env python # -*- coding: utf-8 -*- # # 预测房地产价格 # # author: yafei([email protected]) # import sys import codecs import locale from dtree_loss import DTreeLoss from dtree_parameter import DTreeParameter from dtree_sample import DTreeSample if __name__ == '__main__': param = DTreeParameter() sample = DTreeSample() sample.load('real-estate.txt') dt = DTreeLoss(sample, param) dt.train(None) feature_map = { 0: u'结构', 1: u'装修', 2: u'周边', 3: u'地段', 4: u'绿化', 5: u'交通', 6: u'户均车位', } # 为了输出中文
last_tree = self.trees[i-1]
residual = last_tree.next_residual()
print >>sys.stderr, 'training tree #%d' % (i)
tree.train(residual)
self.trees.append(tree)
def predict(self, x):
y = self.F0
for tree in self.trees:
y += tree.predict(x)
return y
if __name__ == '__main__':
param = DTreeParameter()
param.max_level = 4
param.split_threshold = 0.8
param.max_attr_try_time = 1000
param.tree_number = 20
param.learning_rate = 0.5
sample = DTreeSample()
sample.load_liblinear('heart_scale.txt')
gbdt = GBDT(sample)
gbdt.train(param)
print gbdt.predict([0.708333,1,1,-0.320755,-0.105023,-1,1,-0.419847,-1,-0.225806,0,1,-1])
print gbdt.predict([0.583333,-1,0.333333,-0.603774,1,-1,1,0.358779,-1,-0.483871,0,-1,1])
#! /usr/bin/env python # -*- coding: utf-8 -*- # # 预测weibo粉丝是否是僵尸粉 # # author: yafei([email protected]) # import sys import codecs import locale from dtree_gain import DTreeGain from dtree_parameter import DTreeParameter from dtree_sample import DTreeSample if __name__ == '__main__': param = DTreeParameter() param.split_threshold = 0.93 sample = DTreeSample() sample.load('weibo.txt') dt = DTreeGain(sample, param) dt.train() feature_map = { 0: u'注册天数', 1: u'加V', 2: u'关注', 3: u'粉丝', 4: u'微博', 5: u'收藏', 6: u'互粉', 7: u'共同好友',
last_tree = self.trees[i - 1]
residual = last_tree.next_residual()
print >> sys.stderr, 'training tree #%d' % (i)
tree.train(residual)
self.trees.append(tree)
def predict(self, x):
y = self.F0
for tree in self.trees:
y += tree.predict(x)
return y
if __name__ == '__main__':
param = DTreeParameter()
param.max_level = 4
param.split_threshold = 0.8
param.max_attr_try_time = 1000
param.tree_number = 20
param.learning_rate = 0.5
sample = DTreeSample()
sample.load_liblinear('heart_scale.txt')
gbdt = GBDT(sample)
gbdt.train(param)
print gbdt.predict([
0.708333, 1, 1, -0.320755, -0.105023, -1, 1, -0.419847, -1, -0.225806,
0, 1, -1
])
