notebook.community

Edit and run 





In [1]:



    


import numpy as np
import pandas as pd

import matplotlib.pyplot as plt
%matplotlib inline

Preprocess data





In [2]:



    


# Import data
path = "../data/petdata_1000_100.csv"
raw_data = pd.read_csv(path, index_col="doc_uri")
assert raw_data.shape == (1000,100), "Import error, df has false shape"

Conversion and cleaning

Surprise forces you to use schema ["user_id", "doc_id", "rating"]

CF models are often sensitive to NA values -> replace NaN with 0 (TBD: originally 0 codes to non-rating, but doc already shown to user)





In [3]:



    


# Convert df
data = raw_data.unstack().to_frame().reset_index()
data.columns = ["user", "doc_uri", "rating"]

# Missing value handling
data.fillna(0, inplace=True) 

assert data.shape == (raw_data.shape[0] * raw_data.shape[1], 3), "Conversion error, df has false shape"
assert data.rating.max() <= 5., "Value error, max rating over upper bound"
assert data.rating.min() >= 0., "Value error, min rating under lower bound"
data.head()



    


















    
Out[3]:











  
    

      
      user
      doc_uri
      rating
    

  
  
    

      0
      Aaron Keith III
      http://www.vargas.biz/login.php
      1.0
    

    

      1
      Aaron Keith III
      http://wallace-walker.info/index/
      1.0
    

    

      2
      Aaron Keith III
      http://www.jimenez.biz/
      3.0
    

    

      3
      Aaron Keith III
      http://www.logan.com/about.html
      0.0
    

    

      4
      Aaron Keith III
      http://cox.org/list/tag/faq.html
      5.0

Descriptive statistics of ratings

Not meaningful <- randomly generated





In [4]:



    


data.rating.describe().to_frame().T



    


















    
Out[4]:











  
    

      
      count
      mean
      std
      min
      25%
      50%
      75%
      max
    

  
  
    

      rating
      100000.0
      1.6925
      1.826367
      0.0
      0.0
      1.0
      3.0
      5.0
    

  




















In [5]:



    


# Plot distribution of (random) ratings 

plt.rcParams['figure.figsize'] = [9, 3]

plt.subplot(1, 2, 1)
hist = data.rating.plot(kind="hist", grid=True,
                        bins=[-0.1,0.1,0.9,1.1,1.9,2.1,2.9,3.1,3.9,4.1,4.9,5.1])
hist.set(xlabel= "rating")

plt.subplot(1, 2, 2)
box = data.rating.plot(kind="box")

plt.tight_layout()
plt.savefig("plots/ratings.png", orientation="landscape", dpi=120)

Recommendation Engines

Model-based Collaborative Filtering





In [6]:



    


from surprise import SVD, Dataset, Reader, NMF, accuracy
from surprise.model_selection import cross_validate
from surprise.prediction_algorithms.random_pred import NormalPredictor

reader = Reader(rating_scale=(1, 5))
ds = Dataset.load_from_df(data[["user", "doc_uri", "rating"]], reader)

baseline_model = NormalPredictor() # Baseline model, predicts labels based on distribution of ratings

Matrix factorization-based CF





In [7]:



    


# Models - tune parameters, if you'd like ;)
svd = SVD() # Singular Value Decomposition
nmf = NMF() # Non-negative Matrix factorization

Model Evaluation

Don't expect accurate models <- they are trained with random noise





In [8]:



    


for algo in [baseline_model, svd, nmf]:
    cross_validate(algo, ds, measures=["RMSE", "MAE"], cv=5, verbose=True)



    


















    






Evaluating RMSE, MAE of algorithm NormalPredictor on 5 split(s).

                  Fold 1  Fold 2  Fold 3  Fold 4  Fold 5  Mean    Std     
RMSE (testset)    2.2390  2.2316  2.2146  2.2412  2.2386  2.2330  0.0097  
MAE (testset)     1.8609  1.8607  1.8439  1.8664  1.8672  1.8598  0.0084  
Fit time          0.08    0.12    0.08    0.07    0.07    0.08    0.02    
Test time         0.12    0.12    0.11    0.11    0.10    0.11    0.01    
Evaluating RMSE, MAE of algorithm SVD on 5 split(s).

                  Fold 1  Fold 2  Fold 3  Fold 4  Fold 5  Mean    Std     
RMSE (testset)    1.9133  1.9143  1.9141  1.9208  1.9139  1.9153  0.0028  
MAE (testset)     1.6388  1.6390  1.6402  1.6455  1.6414  1.6410  0.0024  
Fit time          5.37    5.34    5.95    6.06    6.32    5.81    0.39    
Test time         0.13    0.15    0.13    0.14    0.13    0.13    0.01    
Evaluating RMSE, MAE of algorithm NMF on 5 split(s).

                  Fold 1  Fold 2  Fold 3  Fold 4  Fold 5  Mean    Std     
RMSE (testset)    1.8283  1.8277  1.8260  1.8299  1.8351  1.8294  0.0031  
MAE (testset)     1.5751  1.5722  1.5705  1.5768  1.5800  1.5749  0.0033  
Fit time          6.77    6.47    6.91    6.22    6.59    6.59    0.24    
Test time         0.21    0.21    0.12    0.16    0.12    0.16    0.04

k-NN-based CF





In [9]:



    


from surprise.prediction_algorithms.knns import KNNBasic

sim_options = {"name": "pearson", # pearson's r
               "user_based": False  # item-based
               }

knn = KNNBasic(sim_options=sim_options)

Model Evaluation

Don't expect accurate models <- they are trained with random noise





In [10]:



    


cross_validate(knn, ds, measures=["RMSE", "MAE"], cv=5, verbose=True)



    


















    






Computing the pearson similarity matrix...
Computing the pearson similarity matrix...
Computing the pearson similarity matrix...
Computing the pearson similarity matrix...
Computing the pearson similarity matrix...
Done computing similarity matrix.
Done computing similarity matrix.
Done computing similarity matrix.
Done computing similarity matrix.
Done computing similarity matrix.
Evaluating RMSE, MAE of algorithm KNNBasic on 5 split(s).

                  Fold 1  Fold 2  Fold 3  Fold 4  Fold 5  Mean    Std     
RMSE (testset)    1.7801  1.7824  1.7901  1.7829  1.7809  1.7833  0.0035  
MAE (testset)     1.5493  1.5491  1.5542  1.5501  1.5469  1.5499  0.0024  
Fit time          6.21    6.00    6.52    7.74    4.56    6.21    1.02    
Test time         15.85   13.38   15.52   14.39   16.20   15.07   1.04    










    
Out[10]:








{'fit_time': (6.211174488067627,
  5.996851444244385,
  6.519110202789307,
  7.739609956741333,
  4.562095403671265),
 'test_mae': array([1.54929938, 1.54914995, 1.55422394, 1.55005547, 1.54685839]),
 'test_rmse': array([1.78010419, 1.78244229, 1.79006299, 1.78291537, 1.78085411]),
 'test_time': (15.854188680648804,
  13.37714147567749,
  15.522441864013672,
  14.388406753540039,
  16.198522567749023)}

Final model and predictions

kNN Basic looks most promising so far, so go for this one

Train & evaluate final model





In [11]:



    


# Train final model
trainset = ds.build_full_trainset()
knn.fit(trainset)

# RMSE of final model
testset = trainset.build_testset()
test_pred = knn.test(testset)
accuracy.rmse(test_pred, verbose=True) # should be very bad ;)



    


















    






Computing the pearson similarity matrix...
Done computing similarity matrix.
RMSE: 1.3321










    
Out[11]:








1.3321412404968649

Predict some document ratings





In [12]:



    


combinations_to_predict = [("Aaron Keith III", "http://gregory.com/"),
                           ("Abigail Wong", "http://hicks.com/"),
                           ("Julie Bullock", "https://www.garcia.com/"),
                           ("Victoria Perez", "http://lee-phillips.org/register/")]



    











In [13]:



    


# Predictions
for combination in combinations_to_predict:
    user = combination[0]
    doc = combination[1]
    pred = knn.predict(user, doc)
    print(pred[0], "should rate", pred[1], "with", int(round(pred[3])), "stars")



    


















    






Aaron Keith III should rate http://gregory.com/ with 1 stars
Abigail Wong should rate http://hicks.com/ with 1 stars
Julie Bullock should rate https://www.garcia.com/ with 1 stars
Victoria Perez should rate http://lee-phillips.org/register/ with 2 stars

















In [ ]:

Content source: donK23/pyData-Projects

Similar notebooks:

notebook.community | gallery | about