gbdt可视化

gbdt的最大优点，和决策树一样，高度可解释，最喜欢的分类模型：）

#!/usr/bin/env python

#coding=gbk

# ==============================================================================

# file print-fastreank-tree.py

# author chenghuige

# date 2014-10-04 00:34:59.825146

# Description

# ==============================================================================

   

import sys,os

from gflags import *

from gezi import *

from libmelt import *

from BinaryTree import *

from TreeWriter import *

   

DEFINE_string('model', './model/model.json', '')

DEFINE_string('feature', '', '')

DEFINE_integer('tree', 0, '-1 means print all trees')

DEFINE_boolean('use_invisable_node', False, '')

DEFINE_string('outfile', 'tree.png', '')

   

def get_tree_(node_idx, fe, tree, fnames, is_inpath):

node = Node()

node.attr = {'color' : ".7 .3 1.0", 'style' : 'filled'}

node.leftEdgeAttr = {'color' : 'blue', 'penwidth' : '2.5', 'label' : '<='}

node.rightEdgeAttr = {'color' : 'green', 'penwidth' : '2.5', 'label' : '>'}

if is_inpath:

node.attr['color'] = '#40e0d0'

if node_idx < 0:

node.attr['shape'] = 'box'

node.attr['label'] = str(tree.leafValue[-node_idx - 1])

if is_inpath:

print node.attr['label']

return node

name = fnames[tree.splitFeature[node_idx]]

label = '%sl%f <= %f?l[%f]'%(name, fe[tree.splitFeature[node_idx]], tree.threshold[node_idx], tree.previousLeafValue[node_idx])

node.attr['label'] = label

if is_inpath:

l = fe[tree.splitFeature[node_idx]] <= tree.threshold[node_idx]

r = 1 - l

if l:

node.leftEdgeAttr['color'] = 'black'

else:

node.rightEdgeAttr['color'] = 'black'

else:

l = r = 0

node.left = get_tree_(tree.lteChild[node_idx], fe, tree, fnames, l)

node.right = get_tree_(tree.gtChild[node_idx], fe, tree, fnames, r)

return node

   

def get_tree(model, fe, index):

tree = model.trees[index]

fnames = model.Predictor.featureNames

btree        = BinaryTree()

node_idx = 0

btree.root = get_tree_(node_idx, fe, tree, fnames, 1)

return btree        

   

   

def main(argv):

try:

argv = FLAGS(argv) # parse flags

except gflags.FlagsError, e:

print '%s Usage: %s ARGS %s' % (e, sys.argv[0], FLAGS)

sys.exit(1)

   

model = jsonfile2obj(FLAGS.model)

fe = Vector(FLAGS.feature)

tree = get_tree(model, fe, FLAGS.tree)

   

writer = TreeWriter(tree)

if FLAGS.use_invisable_node:

writer.use_invisable_node = True

writer.Write(FLAGS.outfile)

   

if __name__ == "__main__":

main(sys.argv)

   

#!/usr/bin/env python

#coding=gbk

# ==============================================================================

# file TreeWriter.py

# author chenghuige

# date 2014-10-02 20:32:25.744069

# Description

# ==============================================================================

   

import sys

from BinaryTree import *

import pygraphviz as pgv

'''

treeWriter with func wite can write a binary tree to tree.png or user spcified

file

'''

class TreeWriter():

def __init__(self, tree):

self.num = 1 #mark each visible node as its key

self.num2 = -1 #makk each invisible node as its key

self.tree = tree

self.use_invisable_node = False

   

def Write(self, outfile = 'tree.png'):

def writeHelp(root, A):

if not root:

return

   

p = str(self.num)

self.num += 1

A.add_node(p, **root.attr)

q = None

r = None

   

if root.left:

q = writeHelp(root.left, A)

A.add_edge(p, q, **root.leftEdgeAttr)

if root.right:

r = writeHelp(root.right, A)

A.add_edge(p, r, **root.rightEdgeAttr)

   

if not self.use_invisable_node:

return p

   

if q or r:

if not q:

q = str(self.num2)

self.num2 -= 1

A.add_node(q, style = 'invis')

A.add_edge(p, q, style = 'invis')

if not r:

r = str(self.num2)

self.num2 -= 1

A.add_node(r, style = 'invis')

A.add_edge(p, r, style = 'invis')

l = str(self.num2)

self.num2 -= 1

A.add_node(l, style = 'invis')

A.add_edge(p, l, style = 'invis')

B = A.add_subgraph([q, l, r], rank = 'same')

B.add_edge(q, l, style = 'invis')

B.add_edge(l, r, style = 'invis')

   

return p #return key root node

   

self.A = pgv.AGraph(directed=True,strict=True)

writeHelp(self.tree.root, self.A)

self.A.graph_attr['epsilon']='0.001'

#self.A.layout(prog='dot')

#print self.A.string() # print dot file to standard output

self.A.layout('dot') # layout with dot

self.A.draw(outfile) # write to file

   

   

if __name__ == '__main__':

tree = BinaryTree()

tree.CreateTree(-1)

tree.InorderTravel()

writer = TreeWriter(tree)

if len(sys.argv) > 1:

outfile = sys.argv[1]

writer.Write(outfile) #write result to outfile

else:

writer.Write() #write result to tree.png