#!/usr/bin/python
import math

actualFileName = '/home/gregstoll/projects/netflixprize/proberatings.txt'
predictionFileName = '/home/gregstoll/stuff/netflixprize/download/dummyoutput.txt'

def calculateRMSE(aFN, pFN):
    aF = open(aFN, 'r')
    pF = open(pFN, 'r')
    aStr = aF.readline()
    pStr = pF.readline()
    total = 0.0
    count = 0
    while (pStr != ''):
        aStr = aStr.strip()
        pStr = pStr.strip()
        if (aStr[len(aStr)-1:] == ':'):
            if (pStr[len(pStr)-1:] != ':'):
                print "Line mismatch! aStr=%s, pStr=%s" % (aStr, pStr)
        else:
            aNum = float(aStr)
            pNum = float(pStr)
            total = total + (aNum - pNum) * (aNum - pNum)
            count = count + 1
        aStr = aF.readline()
        pStr = pF.readline()
    if (aStr != ''):
        print "Warning: fewer lines in prediction file than actual file!!"
    print "found %d ratings (should be 1407563)" % count
    aF.close()
    pF.close()
    rmse = math.sqrt(total/float(count))
    print "rmse is %f" % rmse

if (__name__ == '__main__'):
    calculateRMSE(actualFileName, predictionFileName)