数据挖掘--购物车 推荐算法

 http://mayuxiang.sinaapp.com/?p=216

 关注另一种数据挖掘算法，这种算法主要用于频繁项集的发现。

熟悉网上购物的同学，一定都了解什么是“购物车” ，或者至少知道在超市中购物所推的购物车吧。亚马逊最早提出了一个算法，就是基于购物车的内容来为用户做出新的推荐。

一个简单的例子来看看什么是A-priori 算法

#!/usr/bin/python
#coding:utf-8

data = [['Cat','and','dog','bites'],
['Yahoo','news','claims','cat','mated','with','a','dog','and','produced',
'viable','offspring'],
['Cat','Killer','likely','is','a','big','dog']
,['Professional','free','advice','on','dog','puppy','training'],
['Cat','and','kitten','training','behavior'],
['Dog','&','Cat','providers','dog','training','in','Engene','Oregon'],
['Dog','and','cat','is','slang','term','used','by','police','officers',
'for','a','male-female','relationship'],['Shop','for','your','show','dog',
'grooming','and','pet','suppliers']]

#wash the data and keep words unique in each row
def delDuplication(data):
	pass  #ignore the implementing process here

#ignore those words which rarely appeared
#Threhold: set the smallest appear times
def supportThreshold(res,threhold=1):
	res2={item:res[item] for item in res if len(res[item]) >= threhold}
	return res2  

def statusCount():
	res = {} #define a dictionary to restore each word and its position
	delDuplication(data) #delete duplication words in data list
        for i in range(0,len(data)):
		for word in data[i]:
			if res.has_key(word) == False:
				res[word]=[]
			res[word].append(i)
	res = supportThreshold(res,3)
	print "一维关联：",res
	return res

def comparePos(posX,posY):
        # compare the two position list to find out
	# if the two word appear in same sentences
	pos =[]
	for p1 in posX:
		for p2 in posY:
			if p1 == p2:
				pos.append(p1)
				break
			elif p2 > p1:
				break
	return pos

def apriori(res):  # apriori aglrithom, form a {res,res} conjunction pair
	hres={}
	for itemX in res:
		for itemY in res:  #compare with all elements behind it
			if itemX!= itemY and not hres.has_key((itemX,itemY)) and not hres.has_key((itemY,itemX)):
				pos = comparePos(res[itemX],res[itemY])
				pair = (itemX,itemY)
				hres[pair] = pos
	hres = supportThreshold(hres,3)
	return hres

if __name__ == "__main__":
	oneDe =statusCount()
	twoDe =apriori(oneDe)
	print "二维关联：",twoDe
	fourDe=apriori(twoDe)
	print "四维关联:",fourDe