In [1]:

    

from __future__ import division
import sys
import numpy as np
import scipy as sp
import pandas as pd
from timeit import default_timer as timer
import matplotlib.pyplot as plt
%matplotlib inline
from os.path import basename

import seaborn as sns
sns.set_style('white')

from polara.recommender.data import RecommenderData, RecommenderDataPositive
from polara.recommender.models import SVDModel, CoffeeModel, NonPersonalized
from polara.evaluation import evaluation_engine as ee
from polara.evaluation.plotting import show_hits, show_hit_rates, show_precision_recall, show_ranking, show_relevance, show_ranking_positivity
from polara.tools.mymedialite.mmlwrapper import MyMediaLiteWrapper
from polara.tools.movielens import get_movielens_data, filter_short_head
from polara.tools.printing import print_frames


    

In [2]:

    

ml_file = "ml-1m.zip"
if sys.platform == 'win32':
    lib_path = 'MyMediaLite-3.11/lib/mymedialite' 
else:
    lib_path = 'MyMediaLite-3.11/bin'
data_folder = 'MyMediaLiteData'

def get_file_name(filepath):
    return ''.join(basename(filepath).split('.')[:-1])


    

In [3]:

    

ml_data = get_movielens_data(local_file=ml_file)
movielens = RecommenderData(ml_data, 'userid', 'movieid', 'rating')
movielens.name = get_file_name(ml_file)


    

In [4]:

    

movielens.holdout_size = 1
movielens.shuffle_data = True
movielens.test_sample = None
movielens.random_holdout = False
movielens.permute_tops = True


    

In [5]:

    

coffee = CoffeeModel(movielens)
coffee.mlrank


    

    
Out[5]:





(13, 10, 2)

In [6]:

    

coffee.build()


    

    




Preparing data
CoFFee model training time: 2.18131835875s

In [7]:

    

v, w = coffee._items_factors, coffee._feedback_factors


    

In [8]:

    

rating_model = movielens
rating_model.index.feedback.T


    

    
Out[8]:






  

    

      

      
0
      
1
      
2
      
3
      
4
    
  
  

    

      
new
      
0
      
1
      
2
      
3
      
4
    
    

      
old
      
1
      
2
      
3
      
4
      
5
    
  

In [9]:

    

pos_rating_idx = rating_model.index.feedback.set_index('old').loc[coffee.switch_positive, 'new']
pos_rating_idx


    

    
Out[9]:





3

In [10]:

    

hit_score = {}
almost_score = {}
fail_score = {}
hidden_pos = {}
rating_diff = {}

num_users = rating_model.test.testset.userid.nunique()

for user_id in rating_model.test.testset.userid.unique():
    user_data = rating_model.test.testset.query('userid==@user_id')
    rating_data = rating_model.index.feedback.set_index('old').loc[user_data.rating.values, 'new'].values
    movies_data = user_data.movieid.values

    user_pref = sp.sparse.coo_matrix((np.ones_like(movies_data), (movies_data, rating_data)), shape = (v.shape[0], w.shape[0]))
    recs = v.dot((v.T.dot(user_pref.A).dot(w)).dot(w.T))
        
    hidden_movie = rating_model.test.evalset.query('userid==@user_id').movieid.iloc[0]
    hidden_rating = rating_model.test.evalset.query('userid==@user_id').rating.iloc[0]
    hidden_rating_idx = rating_model.index.feedback.query('old == @hidden_rating').new.iloc[0]
    
    predicted_rating_idx = recs[hidden_movie, :].argmax()

    if predicted_rating_idx == hidden_rating_idx:
        hit_score[user_id] = 1
    else:
        rating_diff[user_id] = hidden_rating_idx - predicted_rating_idx
        
        is_almost_top = ((predicted_rating_idx >= pos_rating_idx) and (hidden_rating_idx >= pos_rating_idx) or
                         (predicted_rating_idx <  pos_rating_idx) and (hidden_rating_idx <  pos_rating_idx))
        if is_almost_top:
            almost_score[user_id] = 1
        else:
            fail_score[user_id] = 1


    

In [11]:

    

RMSE = np.sqrt(np.sum(rating_diff.values())/num_users)
RMSE


    

    
Out[11]:





0.76718717334194098

In [12]:

    

bingo = sum(hit_score.values())
bingo


    

    
Out[12]:





592

In [13]:

    

positive = sum(almost_score.values())
positive


    

    
Out[13]:





538

In [14]:

    

fail = sum(fail_score.values())
fail


    

    
Out[14]:





78

In [15]:

    

num_users


    

    
Out[15]:





1208

In [16]:

    

bingo / num_users


    

    
Out[16]:





0.4900662251655629

In [17]:

    

positive / num_users


    

    
Out[17]:





0.445364238410596

In [18]:

    

fail / num_users


    

    
Out[18]:





0.06456953642384106