Scribd
Upload
24 views

Lab 3 IA2

The document presents code for building a decision tree classifier using the ID3 algorithm. It defines functions for calculating entropy, information gain, identifying class labels, and recursively constructing the decision tree from a training dataset. The code is then executed on a sample dataset to build and output the decision tree.

Uploaded by

Chad Christensen
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as TXT, PDF, TXT or read online on Scribd
24 views

Lab 3 IA2

The document presents code for building a decision tree classifier using the ID3 algorithm. It defines functions for calculating entropy, information gain, identifying class labels, and recursively constructing the decision tree from a training dataset. The code is then executed on a sample dataset to build and output the decision tree.

Uploaded by

Chad Christensen
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as TXT, PDF, TXT or read online on Scribd
You are on page 1/ 3

import math

from math import log


S = [('age', 'competition', 'type', 'profit'), ('old', 'yes', 'swr', 'down'), ('
old', 'no', 'swr', 'down'), ('old', 'no', 'hwr', 'down'),\
('mid', 'yes', 'swr', 'down'), ('mid', 'yes', 'hwr', 'down'), ('mid', 'no', 'hwr
', 'up'), ('mid', 'no', 'swr', 'up'), ('new', 'yes', 'swr', 'up'), \
('new', 'no', 'hwr', 'up'), ('new', 'no', 'swr', 'up')]
class Tree(object):
def __init__(self):
self.children = None
self.leg = None
self.data = None
def entropy(S):
dicti = {}
for i in range(1, len(S)):
if dicti.get(S[i][3], None) is None:
dicti[S[i][3]] = 1
else:
dicti[S[i][3]] = dicti[S[i][3]] + 1
result = 0
for ke in dicti.keys():
result = result - (dicti[ke]/float(len(S) - 1))*log((dicti[ke]/f
loat(len(S)-1)), 2)
return result
def gain(attrib_pos):
dictiat = {}
for i in range(1, len(S)):
if dictiat.get(S[i][attrib_pos], None) is None:
dictiat[S[i][attrib_pos]] = {}
if dictiat[S[i][attrib_pos]].get(S[i][3], None) is None:
dictiat[S[i][attrib_pos]][S[i][3]] = 1
else:
dictiat[S[i][attrib_pos]][S[i][3]] = dictiat[S[i][attrib
_pos]][S[i][3]] + 1
return dictiat
def getnrcls(S):
dicu = {}
for i in range(1, len(S)):
if dicu.get(S[i][3], None) is None:
dicu[S[i][3]] = 0
return dicu.keys()
def gain_calc(attrib_pos, S):
dic = gain(attrib_pos)
result = entropy(S)
for key in dic.keys():
if len(dic[key].keys()) == 2:
ceva = 0
toate = 0
for keyy in dic[key].keys():
toate = toate + dic[key][keyy]
for keyy in dic[key].keys():
ceva = ceva - dic[key][keyy]/float(toate)*log(di
c[key][keyy]/float(toate),2)
ceva = ceva * toate / (len(S) - 1)
result = result - ceva

return result
dicc = {}
dicc['new'] = 0
dicc['old'] = 0
dicc['mid'] = 0
dicc['yes'] = 1
dicc['no'] = 1
dicc['swr'] = 2
dicc['hwr'] = 2
def id3(S, dicc):
maxx = -10000
for i in range(0, 2):
if maxx < gain_calc(i, S):
maxx = gain_calc(i, S)
pos = i
id3tree = Tree()
id3tree.data = S[0][pos]
id3tree.children = [Tree() for i in range(3)]
id3tree.children[0].leg = 'new'
id3tree.children[1].leg = 'mid'
id3tree.children[2].leg = 'old'
for i in range(len(id3tree.children)):
continueid3(S, id3tree.children[i], id3tree.children[i].leg, dic
c)
return id3tree
def continueid3(S, id3tree, leg, dicc):
newS = [S[0]]
for i in range(1, len(S)):
if S[i][dicc[leg]] == leg:
newS.append(S[i])
if len(getnrcls(newS)) == 1:
id3tree.data=getnrcls(newS)[0]
else:
maxx = -10000
for i in range(1,2):
print gain_calc(i, newS)
if maxx < gain_calc(i, newS):
maxx = gain_calc(i, newS)
pos = i
id3tree.data = newS[0][pos]
id3tree.children = [Tree() for i in range(2)]
id3tree.children[0].leg = 'yes'
id3tree.children[1].leg = 'no'
for i in range(len(id3tree.children)):
continueid3(newS, id3tree.children[i], id3tree.children[
i].leg, dicc)
copac = id3(S, dicc)
print copac.data
print copac.children[0].leg
print copac.children[0].data
print copac.children[1].leg
print copac.children[1].data
print copac.children[2].leg
print copac.children[2].data

print copac.children[1].children[0].leg
print copac.children[1].children[0].data
print copac.children[1].children[1].leg
print copac.children[1].children[1].data
dike = {}
dike['up'] = []
dike['down'] = []
def afis(copac, dic, currentval):
if copac.children is None:
dic[copac.data].append(currentval+copac.leg)
else:
if copac.leg is None:
currentval = currentval + copac.data + ' '
else:
currentval = currentval + copac.leg + ':' + copac.data +
' '
for i in copac.children:
dic = afis(i, dic, currentval)
return dic
print afis(copac, dike, '')

You might also like