Pearson Commit

collab-filter.py

 import csv
-
+from math import sqrt
 csvfile=open('movie-ratings.csv')
-read=csv.reader(csvfile)
-reading=[]
-for line in read:
-	reading.append(line)
+reading=list(csv.reader(csvfile))
+#########################
 f2 = open ("user_preference.csv")
-data = csv.reader(f2)
+user_data = csv.reader(f2)
 preference = []
-a=input("Your name :")
-preference.append(a)
-for line in data:
+preference.append(input("Your name :"))
+for line in user_data:
     preference.append(line)
 prefdict = dict( zip(preference[1],preference[2]))
-print(prefdict)
\ No newline at end of file
+####################
+header = reading[0]
+del reading[0]
+reviewers = [ s[0] for s in reading]
+reviewdic = []
+for line in reading:
+    dic = dict( (header[i],line[i]) for i in range(1,len(line)))
+    reviewdic.append(dic)
+dist=[]
+for i in reviewdic:
+    sum1=0
+    euclid_temp=[sqrt((float(i[j])-float(prefdict[j]))**2) for j in prefdict.keys() if float(i[j])!=-1 and float(prefdict[j])!=-1]
+    dist.append(sum(euclid_temp)/len(euclid_temp))
+order_dist = sorted(zip(reviewers,dist), key=lambda x: x[1],reverse=True)
+#############################################
+temp_movieUser=[i for i in prefdict.keys() if float(prefdict[i])!=-1]
+pear=[]
+for i in reviewdic:
+    temp_movieRev=[j for j in prefdict.keys() if float(i[j])!=-1]
+    var=list(set(temp_movieUser).intersection(temp_movieRev))
+    temp=[float(i[j]) for j in var if float(i[j])!=-1]
+    temp1=[float(prefdict[j]) for j in var if float(prefdict[j])!=-1]
+    mean=sum(temp)/len(temp)
+    avgUser=sum(temp1)/len(temp1)
+    stdUser=sqrt((sum(map(lambda x: x*x,temp1))/len(temp1))-avgUser)
+    stdRev=sqrt((sum(map(lambda x: x*x,temp))/len(temp))-mean)
+    pear.append(sum([(float(prefdict[j])-avgUser)*(float(i[j])-mean)/(len(var)*stdRev*stdUser) for j in var]))
+order_pear=sorted(zip(reviewers,pear), key=lambda x:x[1], reverse=True)

user_preference.csv

-The Godfather,Pulp Fiction,The Clockwork Orange,Cinema Paradiso,Wild Strawberries,Amadeus
-9,7,6,10,8,10
+The Godfather,Pulp Fiction,Gone With the Wind,12 Angry Men,Lawrence of Arabia,Raging Bull,The Clockwork Orange,Cinema Paradiso,Wild Strawberries,Amadeus
+9,7,6,10,8,10,-1,-1,-1,-1