#!/usr/bin/env python
# encoding: utf-8
"""
learning.py

Created by Cody Mays on 2009-10-20.
"""

import sys
import os
import random
from mldata import *

# Global vars
moviesAvg = [1.0] * 30
userAvg = [1.0] * len(allRatings)

def task1():
	total = 0
	divisor = 0
	
	# Movie average
	print 'Movie averages:\n'
	for i in range(0,len(allRatings[0])):
		movieTotal = 0
		movieDivisor = 0
		
		for j in range(0, len(allRatings)):
			x = allRatings[j]
			y = allRatings[j][i]
			
			if(allRatings[j][i] != 0):
				movieTotal += allRatings[j][i]
				movieDivisor += 1
		
		if(movieDivisor == 0):
			moviesAvg[i] = 0
		else:
			moviesAvg[i] = movieTotal/movieDivisor
		
		print '#',i,':',moviesAvg[i]
	
	# User average...
	print ''
	print 'User averages: '
	for i in range(0, len(allRatings)):
		for j in range(0, len(allRatings[i])):
			if(allRatings[i][j] != 0):
				total += allRatings[i][j]
				divisor += 1
		
		# Make sure divisor isn't 0 and output the average for each user
		if(divisor != 0):
			userAvg[i] = total/divisor
		else:
			userAvg[i] = 0
			
		print names[i],':',userAvg[i]
			
		total = divisor = 0

def task2():
	correlationScores = [0 for i in range(len(allRatings))]
	normVals = [[0 for col in range( (len(allRatings[0])+1))] for row in range(len(allRatings))]
	dotProducts = [0 for i in range(len(allRatings))]
	maxCV = 0
	me = 16
	closest = 0
	
	# Normalize scores vector and store
	for user in range(0, len(allRatings)):
		for j in range(0, len(allRatings[user])):
			normVals[user][j] = allRatings[user][j] - userAvg[user]
	
	# Find closest correlation
	for user in range(0, len(allRatings)):
		dotProd = 0
		if(user != me):
			#calc dot product
			for j in range(0, len(allRatings[user])):
				dotProd += normVals[me][j] * normVals[user][j]
			dotProducts[user] = dotProd

			if(dotProducts[user] > dotProducts[maxCV]):
				maxCV = user
		
	# Try to find closest user
	for user in range(0, len(allRatings)):
		if( abs(dotProducts[me] - dotProducts[closest]) > abs(dotProducts[me] - dotProducts[user]) ):
			closest = user
	
	print "The person with the highest C is", names[maxCV]
	print "The closest user to you is", names[user]
 
def update(user, movie, actualRating):
       predictedRating = moviesAvg[movie] * userAvg[user]
       error = actualRating - predictedRating
       userAvg[user] += error * moviesAvg[movie] * 0.0001
       moviesAvg[movie] += error * userAvg[user] * 0.0001

def task3():
	maxMovie = 0
	minMovie = 1
	maxUser = 0
	minUser = 1
	
	
	for i in range(0,500):
		movie = random.randint(0, (len(allRatings[0])-1))
		user = random.randint(0, (len(allRatings)-1))
		update(user, movie, allRatings[user][movie])
		
		if(moviesAvg[movie] > moviesAvg[maxMovie]):
			maxMovie = movie
		elif(moviesAvg[movie] < moviesAvg[minMovie]):
			minMovie = movie
			
		if(userAvg[user] > userAvg[maxUser]):
			maxUser = user
		elif(userAvg[user] < userAvg[minUser]):
			minUser = user
	
	print 'Most influential movie:', maxMovie
	print 'Least influential movie:', minMovie
	print 'Most influenced user:',maxUser,names[maxUser]
	print 'Least influenced user:',minUser,names[minUser]
	
	# Predict my movie
	# I'm id #16
	myMovie = 0
	for i in range(0, len(moviesAvg)):
		r = userAvg[16] * moviesAvg[i]
		
		if(r > (userAvg[myMovie] * moviesAvg[myMovie])):
			myMovie = i
	print 'I should watch movie', myMovie

def main():
	task1()
	task2()
	task3()

if __name__ == '__main__':
	main()