from math260 import data_prep, recommend, score
import matplotlib.pyplot as plt
import numpy as np
from tqdm import tqdm
# choose your datasets
GAMES_DATA_FILE = "data/games.csv"
REVIEWS_FILE = "data/reviews.csv"
games, users = data_prep.parse_data(GAMES_DATA_FILE, REVIEWS_FILE, True)
games_map, users_map, rating_matrix, bool_matrix = \
data_prep.create_review_matrix(games, users, sparse=False, verbose=True)
# creates a bar plot of the frequency of each rating
score_counts = np.zeros(11)
game_names = list(games.keys())
print('Iterating over all reviews')
for name in tqdm(game_names):
game = games[name]
reviews = game['reviews']
for review in reviews:
score_counts[int(review['rating'])] += 1
plt.bar(np.arange(11), score_counts)
plt.xlabel('Rating')
plt.ylabel('Number of Reviews')
plt.title('Distribution of Ratings')
plt.savefig('figures/rating-dist.png')
from math260 import data_prep, recommend, score
import numpy as np
GAMES_FILE = "data/games.csv"
REVIEWS_FILE = "data/reviews.csv"
if __name__ == "__main__":
games, users = data_prep.parse_data(GAMES_FILE, REVIEWS_FILE, verbose=True)
games_map, users_map, rating_matrix, bool_matrix \
= data_prep.create_review_matrix(games, users, sparse=False, verbose=True)
game_predictor = recommend.AveragePredictor(np.copy(rating_matrix),
np.copy(bool_matrix))
removed = score.remove_fraction(0.1, rating_matrix, bool_matrix)
# testing removing 10% from each user and predicting using average score
global_predictor = recommend.GlobalAveragePredictor(
rating_matrix, bool_matrix)
user_predictor = recommend.UserAveragePredictor()
tw_predictor = recommend.TwoWayAveragePredictor(rating_matrix, bool_matrix)
glob_avg_rmse = score.rmse(removed,
rating_matrix,
bool_matrix,
global_predictor.predict,
users=range(0, 1000))
game_avg_rmse = score.rmse(removed,
rating_matrix,
bool_matrix,
game_predictor.predict,
users=range(0, 1000))