(Recommendation Systems) Problem Definition

Lets say we have 8 users, and they have rated 8 different albums on a scale of 1 to 5. Note that not all users have rated all albums. songData3 = {"Angelica": {"Blues Traveler": 3.5, "Broken Bells": 2.0, "Norah Jones": 4.5, "Phoenix": 5.0, "Slightly Stoopid": 1.5, "The Strokes": 2.5, "Vampire Weekend": 2.0}, "Bill":{"Blues Traveler": 2.0, Broken Bells": 3.5, "Deadmau5": 4.0, "Phoenix": 2.0, "Slightly Stoopid": 3.5, "Vampire Weekend": 3.0}, "Chan": {"Blues Traveler": 5.0, "Broken Bells": 1.0, "Deadmau5": 1.0, "Norah Jones": 3.0, "Phoenix": 5, "Slightly Stoopid": 1.0}, "Dan": {"Blues Traveler": 3.0, "Broken Bells": 4.0, "Deadmau5": 4.5, "Phoenix": 3.0, "Slightly Stoopid": 4.5, "The Strokes": 4.0, "Vampire Weekend": 2.0}, "Hailey": {"Broken Bells": 4.0, "Deadmau5": 1.0, "Norah Jones": 4.0, "The Strokes": 4.0, "Vampire Weekend": 1.0}, "Jordyn": {"Broken Bells": 4.5, "Deadmau5": 4.0, "Norah Jones": 5.0, "Phoenix": 5.0, "Slightly Stoopid": 4.5, "The Strokes": 4.0, "Vampire Weekend": 4.0}, "Sam": {"Blues Traveler": 5.0, "Broken Bells": 2.0, "Norah Jones": 3.0, "Phoenix": 5.0, "Slightly Stoopid": 4.0, "The Strokes": 5.0}, "Veronica": {"Blues Traveler": 3.0, "Norah Jones": 5.0, "Phoenix": 4.0, "Slightly Stoopid": 2.5, "The Strokes": 3.0}}

Requirement for this Assignment You have been provided with a framework for this assignment: Assignment-3-Framework.py. The Framework defines a Class called similarity. You must use the class as-is with no changes. The class includes a class initialization method that takes two rating dictionaries ratingP and ratingQ as parameters. The class has two methods:

minkowski which takes a single parameter r, and returns the Minkowki Distance between the two dictionaries (that the Class object is instantiated with).

pearson which takes no parameters, and returns the Pearson Correlation between the two dictionaries (that the Class object is instantiated with).

Given a userX, write code to find the NN (k=1) recommendations for userX based on the Euclidean Similarity measure. Pseudo-code has been provided to you in the framework. So, you essentially need to plug in appropriate code for steps 1 through 5 in the framework. Since this is the same data as used in the Recommendation Systems Lecture, you should check your output against the Lecture for accuracy. As covered during our lecture, you may want to review the Assignment-3-Hints.py script with general Python statements to understand how to navigate data structures for this assignment. Assignment Submission Some things to keep in mind as you code:

Make your code readable for instance, use meaningful variable names and comments.

Make your code elegant for instance, balance the number of variables you introduce too many or too few make your code difficult to debug, read, and maintain.

Make your output readable and user-friendly Once you have written up the script, save it as follows. Submit the script by uploading your python script. Note: upload the actual script DO NOT attach a screenshot of the script! Assignment3.py. The submitted script will be run as-is for grading using the Anaconda Spyder 3.7.0 installation required for this class.

Points will be deducted for scripts that:

are difficult to read/follow

dont compile/run

dont have all the various pieces of code required

have hard-code values instead of using variables

have logical errors dont result in the expected output

dont have user-friendly output

dont run efficiently following the practices and lecture examples demonstrated in our lectures and problem sets (located in the course site)

FRAMEWORK:

Page

# -*- coding: utf-8 -*-"""Author Name : Insert Your Name HereDate: Insert the Date """print ()import mathfrom operator import itemgetter# define class similarityclass similarity: # Class instantiation def __init__ (self, ratingP, ratingQ): self.ratings1 = ratingP self.ratings2 = ratingQ # Minkowski Distance between two vectors def minkowksi(self, r): # calculate minkowski distance distance = 0 for k in (set(self.ratings1.keys()) & set(self.ratings2.keys())): p = self.ratings1[k] q = self.ratings2[k] distance += pow(abs(p - q), r) # return value of minkowski distance return pow(distance,1/r) # Pearson Correlation between two vectors def pearson(self): sumpq = 0 sump = 0 sumq = 0 sump2 = 0 sumq2 = 0 n = 0 # calculate pearson correlation using the computationally efficient form for k in (set(self.ratings1.keys()) & set(self.ratings2.keys())): n += 1 p = self.ratings1[k] q = self.ratings2[k] sumpq += p * q sump += p sumq += q sump2 += pow(p, 2) sumq2 += pow(q, 2) # error check for n==0 condition if n == 0: print (">>> pearson debug: n=0; returning -2 correlation!") return -2

# calculate nr and dr for pearson correlation nr = (sumpq - (sump * sumq) / n) dr = (math.sqrt(sump2 - pow(sump, 2) / n) * math.sqrt(sumq2 - pow(sumq, 2) / n)) # error check for dr==0 condition if dr == 0: print (">>> pearson debug: denominator=0; returning -2 correlation!") return -2 # return value of pearson correlation coefficient return nr / dr# user ratingssongData3 = {"Angelica": {"Blues Traveler": 3.5, "Broken Bells": 2.0, "Norah Jones": 4.5, "Phoenix": 5.0, "Slightly Stoopid": 1.5, "The Strokes": 2.5, "Vampire Weekend": 2.0}, "Bill":{"Blues Traveler": 2.0, "Broken Bells": 3.5, "Deadmau5": 4.0, "Phoenix": 2.0, "Slightly Stoopid": 3.5, "Vampire Weekend": 3.0}, "Chan": {"Blues Traveler": 5.0, "Broken Bells": 1.0, "Deadmau5": 1.0, "Norah Jones": 3.0, "Phoenix": 5, "Slightly Stoopid": 1.0}, "Dan": {"Blues Traveler": 3.0, "Broken Bells": 4.0, "Deadmau5": 4.5, "Phoenix": 3.0, "Slightly Stoopid": 4.5, "The Strokes": 4.0, "Vampire Weekend": 2.0}, "Hailey": {"Broken Bells": 4.0, "Deadmau5": 1.0, "Norah Jones": 4.0, "The Strokes": 4.0, "Vampire Weekend": 1.0}, "Jordyn": {"Broken Bells": 4.5, "Deadmau5": 4.0, "Norah Jones": 5.0, "Phoenix": 5.0, "Slightly Stoopid": 4.5, "The Strokes": 4.0, "Vampire Weekend": 4.0}, "Sam": {"Blues Traveler": 5.0, "Broken Bells": 2.0, "Norah Jones": 3.0, "Phoenix": 5.0, "Slightly Stoopid": 4.0, "The Strokes": 5.0}, "Veronica": {"Blues Traveler": 3.0, "Norah Jones": 5.0, "Phoenix": 4.0, "Slightly Stoopid": 2.5, "The Strokes": 3.0} }# for whom are we making recommendations?userX = "Hailey"userXRatings = songData3[userX]# find the euclidean distance between userX's ratings, and each of the other user'sratings.# use a for loop to get at the other users and their ratings - DO NOT hard code.# use the similarity class to caclulate the euclidean distance between user ratings.# assign list of (user, distance) tuples to variable userDistances.# Example: [('Jordyn', 4.39), ('Chan', 3.16), ('Veronica', 1.41), ('Bill', 3.64)]userDistances = []# <<<<< (1) YOUR CODE HERE >>>>># sort list of tuples by lowest distance to highest distance.# assign sorted list to variable userSortedDistances.# Example: [('Veronica', 1.41), ('Sam', 2.45), ('Angelica', 2.74)]userSortedDistances = []# <<<<< (2) YOUR CODE HERE >>>>># userX's NN is the user at the 0th position of the sorted list.# assign NN to userXNN.# Example: 'Veronica'

userXNN = ""# <<<<< (3) YOUR CODE HERE >>>>># recos for userX will include albums rated by userXNN, not already rated by userX.# assign list of (album, rating) tuples to variable userXRecos.# Example: [('Slightly Stoopid', 2.5), ('Blues Traveler', 3.0), ('Phoenix', 4.0)]userXRecos = []# <<<<< (4) YOUR CODE HERE >>>>># sort list of tuples by highest rating to lowest rating.# assign sorted list to varaible userXSortedRecos.# Example: [('Phoenix', 4.0), ('Blues Traveler', 3.0), ('Slightly Stoopid', 2.5)]userXSortedRecos = []# <<<<< (5) YOUR CODE HERE >>>>>print ("Recommendations for", userX)print ("--------------------------")print ()print (userXSortedRecos)