Movies ratings and rankings example

Packages to load

[1]:

import ratingslib.ratings as rl
import pandas as pd
from ratingslib.datasets.filenames import FILENAME_MOVIES_EXAMPLE, dataset_path
from ratingslib.datasets.parse import create_pairs_data, parse_pairs_data
from ratingslib.utils.methods import parse_columns

Prepare data

[2]:

filename = dataset_path(FILENAME_MOVIES_EXAMPLE)
user_movie_df = pd.read_csv(filename, index_col='User')

Create movie-movie dataframe. Movies are placed in pairs which means that every pair is a hypothetical matchup. The columns of movie_movie dataframe are set in COLUMNS dictionary.

[3]:

COLUMNS = {'item_i': 'MovieI',
           'item_j': 'MovieJ',
           'points_i': 'RatingI',
           'points_j': 'RatingJ'
           }

col_names = parse_columns(COLUMNS)
movie_movie_df = create_pairs_data(user_movie_df, columns_dict=COLUMNS)

Parse movie-movie dataframe as pairs data.

[4]:

data_df, items_df = parse_pairs_data(movie_movie_df, columns_dict=COLUMNS)

Colley method

[5]:

rl.Colley().rate(data_df, items_df,
                          columns_dict=COLUMNS, sort=True)

[5]:

	Item	rating	ranking
2	Movie3	0.750	1
1	Movie2	0.425	2
0	Movie1	0.325	3

Massey method

[6]:

rl.Massey().rate(data_df, items_df,
                          columns_dict=COLUMNS, sort=True)

[6]:

	Item	rating	ranking
2	Movie3	1.166667	1
1	Movie2	0.104167	2
0	Movie1	-1.270833	3

Keener method

[7]:

rl.Keener(normalization=True).rate(data_df, items_df,
                                            columns_dict=COLUMNS,
                                            sort=True)

[7]:

	Item	rating	ranking
2	Movie3	0.510552	1
1	Movie2	0.288870	2
0	Movie1	0.200577	3

Offense-Defense method

[8]:

rl.OffenseDefense(tol=0.001).rate(data_df, items_df,
                                       columns_dict=COLUMNS, sort=True)

[8]:

	Item	rating	ranking
2	Movie3	21.998853	1
1	Movie2	15.821345	2
0	Movie1	10.618677	3