Skip to content
Extraits de code Groupes Projets
Valider 4c44fb82 rédigé par Adrien Payen's avatar Adrien Payen
Parcourir les fichiers

update

parent a7471cdc
Aucune branche associée trouvée
Aucune étiquette associée trouvée
Aucune requête de fusion associée trouvée
Masquer les modifications d'espaces
En ligne Côte à côte
Affichage de
avec 2 ajouts et 1 suppression
content_based.ipynb
+ 2
− 1
Voir le fichier @ 4c44fb82
%% Cell type:markdown id:82d5ca82 tags:
# Packages
%% Cell type:code id:277473a3 tags:
``` python
%load_ext autoreload
%autoreload 2
import numpy as np
import pandas as pd
import random as rd
from surprise import AlgoBase
from surprise.prediction_algorithms.predictions import PredictionImpossible
from loaders import load_ratings
from loaders import load_items
from constants import Constant as C
from sklearn.linear_model import LinearRegression
from sklearn.ensemble import GradientBoostingRegressor, RandomForestRegressor
from sklearn.svm import SVR
from sklearn.feature_extraction.text import TfidfVectorizer
```
%% Output
The autoreload extension is already loaded. To reload it, use:
%reload_ext autoreload
%% Cell type:markdown id:a42c16bf tags:
# Explore and select content features
%% Cell type:code id:e8378976 tags:
``` python
# All the dataframes
df_items = load_items()
df_ratings = load_ratings()
df_tag = pd.read_csv(C.CONTENT_PATH/C.TAGS_FILENAME)
df_genome_score = pd.read_csv("data/hackathon/content/genome-scores.csv")
df_genome_tag = pd.read_csv("data/hackathon/content/genome-tags.csv")
# Example 1 : create title_length features
df_features = df_items[C.LABEL_COL].apply(lambda x: len(x)).to_frame('n_character_title')
display(df_features.head())
df_tag = pd.read_csv(C.CONTENT_PATH/C.TAGS_FILENAME)
df_features = df_tag[C.TAG]
display(df_features.head())
# (explore here other features)
```
%% Output
%% Cell type:markdown id:a2c9a2b6 tags:
# Build a content-based model
When ready, move the following class in the *models.py* script
%% Cell type:code id:16b0a602 tags:
``` python
class ContentBased(AlgoBase):
def __init__(self, features_method, regressor_method):
AlgoBase.__init__(self)
self.regressor_method = regressor_method
self.content_features = self.create_content_features(features_method)
self.user_profile_explain = {}
def create_content_features(self, features_method):
"""Content Analyzer"""
df_items = load_items()
df_ratings = load_ratings()
df_tag = df_tag = pd.read_csv(C.CONTENT_PATH/C.TAGS_FILENAME)
df_genome_score = pd.read_csv("data/hackathon/content/genome-scores.csv")
df_genome_tag = pd.read_csv("data/hackathon/content/genome-tags.csv")
if features_method is None:
df_features = None
elif features_method == "relevance" :
df_features = df_genome_score.groupby('movieId')["relevance"].transform('mean').to_frame('avg_relevance')
elif features_method == "title_length": # a naive method that creates only 1 feature based on title length
df_features = df_items[C.LABEL_COL].apply(lambda x: len(x)).to_frame('n_character_title')
elif features_method == "movie_year" :
df_features = df_items['movie_year'] = df_items['title'].str.extract(r'\((\d{4})\)', expand=False).to_frame('movie_year')
elif features_method == "genres" :
genres_list = df_items['genres'].str.split('|').explode().unique()
for genre in genres_list:
df_features = df_items['genres'].str.contains(genre).astype(int).to_frame('genres')
elif features_method == "combination":
df_length = df_items[C.LABEL_COL].apply(lambda x: len(x)).to_frame('n_character_title')
df_movie = df_items['title'].str.extract(r'\((\d{4})\)', expand=False).to_frame('movie_year')
genres_list = df_items['genres'].str.split('|').explode().unique()
for genre in genres_list:
df_genre = df_items['genres'].str.contains(genre).astype(int).to_frame('genres')
df_features = pd.concat([df_genre, df_length, df_movie], axis=1)
elif features_method == "rating" :
df_features = df_ratings.groupby('movieId')['rating'].transform('mean').to_frame('avg_rating')
elif features_method == "tags" :
df_features = df_tag['tag'].apply(lambda x: len(x.split(',')) if isinstance(x, str) else 0).to_frame('tags')
elif features_method == "tags_length" :
df_features = df_tag['tag'].apply(lambda x: sum(len(tag) for tag in x.split(','))if isinstance(x, str) else 0).to_frame('n_character_tags')
else: # (implement other feature creations here)
raise NotImplementedError(f'Feature method {features_method} not yet implemented')
return df_features
def fit(self, trainset):
"""Profile Learner"""
AlgoBase.fit(self, trainset)
# Preallocate user profiles
self.user_profile = {u: None for u in trainset.all_users()}
self.user_profile_explain = {u: {} for u in trainset.all_users()}
for u in self.user_profile :
user_ratings = np.array([rating for _, rating in trainset.ur[u]])
feature_values = self.content_features.values
weighted_features = feature_values.T.dot(user_ratings)
feature_importance = weighted_features / np.sum(user_ratings)
self.user_profile_explain[u] = dict(zip(self.content_features.columns, feature_importance))
if self.regressor_method == 'random_score':
for u in self.user_profile :
self.user_profile[u] = rd.uniform(0.5,5)
elif self.regressor_method == 'random_sample':
for u in self.user_profile:
self.user_profile[u] = [rating for _, rating in self.trainset.ur[u]]
elif self.regressor_method == 'linear_regression' :
for u in self.user_profile:
user_ratings = [rating for _, rating in trainset.ur[u]]
item_ids = [iid for iid, _ in trainset.ur[u]]
df_user = pd.DataFrame({'item_id': item_ids, 'user_ratings': user_ratings})
df_user["item_id"] = df_user["item_id"].map(trainset.to_raw_iid)
df_user = df_user.merge(self.content_features, left_on = "item_id", right_index = True, how = 'left')
if 'n_character_title' in df_user.columns:
# Si 'n_character_title' est disponible comme caractéristique
X = df_user['n_character_title'].values.reshape(-1, 1)
elif 'avg_relevance' in df_user.columns:
# Si 'n_character_title' est disponible comme caractéristique
X = df_user['avg_relevance'].values.reshape(-1, 1)
elif 'movie_year' in df_user.columns:
# Si 'n_character_title' est disponible comme caractéristique
X = df_user['movie_year'].values.reshape(-1, 1)
elif 'genres' in df_user.columns:
# Si 'n_character_title' est disponible comme caractéristique
X = df_user['genres'].values.reshape(-1, 1)
elif 'combination' in df_user.columns:
# Si 'n_character_title' est disponible comme caractéristique
X = df_user['combination'].values.reshape(-1, 1)
elif 'avg_rating' in df_user.columns:
# Si 'n_character_title' est disponible comme caractéristique
X = df_user['avg_rating'].values.reshape(-1, 1)
elif 'tags' in df_user.columns:
# Si une autre caractéristique est disponible (remplace 'other_feature' par le nom de ta caractéristique)
X = df_user['tags'].values.reshape(-1, 1)
elif 'n_character_tags' in df_user.columns:
# Si une autre caractéristique est disponible (remplace 'other_feature' par le nom de ta caractéristique)
X = df_user['n_character_tags'].values.reshape(-1, 1)
else:
# Si aucune caractéristique appropriée n'est disponible
continue # Ou gère le cas d'erreur/exception ici
y = df_user['user_ratings'].values
linear_regressor = LinearRegression(fit_intercept = False)
linear_regressor.fit(X,y)
# Store the computed user profile
self.user_profile[u] = linear_regressor
elif self.regressor_method == 'svr_regression':
for u in self.user_profile:
user_ratings = [rating for _, rating in trainset.ur[u]]
item_ids = [iid for iid, _ in trainset.ur[u]]
df_user = pd.DataFrame({'item_id': item_ids, 'user_ratings': user_ratings})
df_user["item_id"] = df_user["item_id"].map(trainset.to_raw_iid)
df_user = df_user.merge(self.content_features, left_on = "item_id", right_index = True, how = 'left')
if 'n_character_title' in df_user.columns:
# Si 'n_character_title' est disponible comme caractéristique
X = df_user['n_character_title'].values.reshape(-1, 1)
elif 'avg_relevance' in df_user.columns:
# Si 'n_character_title' est disponible comme caractéristique
X = df_user['avg_relevance'].values.reshape(-1, 1)
elif 'movie_year' in df_user.columns:
# Si 'n_character_title' est disponible comme caractéristique
X = df_user['movie_year'].values.reshape(-1, 1)
elif 'genres' in df_user.columns:
# Si 'n_character_title' est disponible comme caractéristique
X = df_user['genres'].values.reshape(-1, 1)
elif 'combination' in df_user.columns:
# Si 'n_character_title' est disponible comme caractéristique
X = df_user['combination'].values.reshape(-1, 1)
elif 'avg_rating' in df_user.columns:
# Si 'n_character_title' est disponible comme caractéristique
X = df_user['avg_rating'].values.reshape(-1, 1)
elif 'tags' in df_user.columns:
# Si une autre caractéristique est disponible (remplace 'other_feature' par le nom de ta caractéristique)
X = df_user['tags'].values.reshape(-1, 1)
elif 'n_character_tags' in df_user.columns:
# Si une autre caractéristique est disponible (remplace 'other_feature' par le nom de ta caractéristique)
X = df_user['n_character_tags'].values.reshape(-1, 1)
else:
# Si aucune caractéristique appropriée n'est disponible
continue # Ou gère le cas d'erreur/exception ici
y = df_user['user_ratings'].values
svr_regressor = SVR(kernel='rbf', C=10, epsilon=0.2)
svr_regressor.fit(X, y)
self.user_profile[u] = svr_regressor
elif self.regressor_method == 'gradient_boosting':
for u in self.user_profile:
user_ratings = [rating for _, rating in trainset.ur[u]]
item_ids = [iid for iid, _ in trainset.ur[u]]
df_user = pd.DataFrame({'item_id': item_ids, 'user_ratings': user_ratings})
df_user["item_id"] = df_user["item_id"].map(trainset.to_raw_iid)
df_user = df_user.merge(self.content_features, left_on = "item_id", right_index = True, how = 'left')
if 'n_character_title' in df_user.columns:
# Si 'n_character_title' est disponible comme caractéristique
X = df_user['n_character_title'].values.reshape(-1, 1)
elif 'avg_relevance' in df_user.columns:
# Si 'n_character_title' est disponible comme caractéristique
X = df_user['avg_relevance'].values.reshape(-1, 1)
elif 'movie_year' in df_user.columns:
# Si 'n_character_title' est disponible comme caractéristique
X = df_user['movie_year'].values.reshape(-1, 1)
elif 'genres' in df_user.columns:
# Si 'n_character_title' est disponible comme caractéristique
X = df_user['genres'].values.reshape(-1, 1)
elif 'combination' in df_user.columns:
# Si 'n_character_title' est disponible comme caractéristique
X = df_user['combination'].values.reshape(-1, 1)
elif 'avg_rating' in df_user.columns:
# Si 'n_character_title' est disponible comme caractéristique
X = df_user['avg_rating'].values.reshape(-1, 1)
elif 'tags' in df_user.columns:
# Si une autre caractéristique est disponible (remplace 'other_feature' par le nom de ta caractéristique)
X = df_user['tags'].values.reshape(-1, 1)
elif 'n_character_tags' in df_user.columns:
# Si une autre caractéristique est disponible (remplace 'other_feature' par le nom de ta caractéristique)
X = df_user['n_character_tags'].values.reshape(-1, 1)
else:
# Si aucune caractéristique appropriée n'est disponible
continue # Ou gère le cas d'erreur/exception ici
y = df_user['user_ratings'].values
gb_regressor = GradientBoostingRegressor(n_estimators=100, learning_rate=0.1, max_depth=3)
gb_regressor.fit(X, y)
self.user_profile[u] = gb_regressor
elif self.regressor_method == 'random_forest':
for u in self.user_profile:
user_ratings = [rating for _, rating in trainset.ur[u]]
item_ids = [iid for iid, _ in trainset.ur[u]]
df_user = pd.DataFrame({'item_id': item_ids, 'user_ratings': user_ratings})
df_user["item_id"] = df_user["item_id"].map(trainset.to_raw_iid)
df_user = df_user.merge(self.content_features, left_on = "item_id", right_index = True, how = 'left')
if 'n_character_title' in df_user.columns:
# Si 'n_character_title' est disponible comme caractéristique
X = df_user['n_character_title'].values.reshape(-1, 1)
elif 'avg_relevance' in df_user.columns:
# Si 'n_character_title' est disponible comme caractéristique
X = df_user['avg_relevance'].values.reshape(-1, 1)
elif 'movie_year' in df_user.columns:
# Si 'n_character_title' est disponible comme caractéristique
X = df_user['movie_year'].values.reshape(-1, 1)
elif 'genres' in df_user.columns:
# Si 'n_character_title' est disponible comme caractéristique
X = df_user['genres'].values.reshape(-1, 1)
elif 'combination' in df_user.columns:
# Si 'n_character_title' est disponible comme caractéristique
X = df_user['combination'].values.reshape(-1, 1)
elif 'avg_rating' in df_user.columns:
# Si 'n_character_title' est disponible comme caractéristique
X = df_user['avg_rating'].values.reshape(-1, 1)
elif 'tags' in df_user.columns:
# Si une autre caractéristique est disponible (remplace 'other_feature' par le nom de ta caractéristique)
X = df_user['tags'].values.reshape(-1, 1)
elif 'n_character_tags' in df_user.columns:
# Si une autre caractéristique est disponible (remplace 'other_feature' par le nom de ta caractéristique)
X = df_user['n_character_tags'].values.reshape(-1, 1)
else:
# Si aucune caractéristique appropriée n'est disponible
continue # Ou gère le cas d'erreur/exception ici
y = df_user['user_ratings'].values
rf_regressor = RandomForestRegressor(n_estimators=100)
rf_regressor.fit(X, y)
self.user_profile[u] = rf_regressor
else :
pass
# (implement here the regressor fitting)
def estimate(self, u, i):
"""Scoring component used for item filtering"""
# First, handle cases for unknown users and items
if not (self.trainset.knows_user(u) and self.trainset.knows_item(i)):
raise PredictionImpossible('User and/or item is unkown.')
if self.regressor_method == 'random_score':
rd.seed()
score = rd.uniform(0.5,5)
elif self.regressor_method == 'random_sample':
rd.seed()
score = rd.choice(self.user_profile[u])
elif self.regressor_method == 'linear_regression':
raw_item_id = self.trainset.to_raw_iid(i)
item_features = self.content_features.loc[raw_item_id:raw_item_id, :].values
linear_regressor = self.user_profile[u]
score= linear_regressor.predict(item_features)[0]
elif self.regressor_method == 'svr_regression':
raw_item_id = self.trainset.to_raw_iid(i)
item_features = self.content_features.loc[raw_item_id:raw_item_id, :].values
svr_regressor = self.user_profile[u]
score = svr_regressor.predict(item_features)[0]
elif self.regressor_method == 'gradient_boosting':
raw_item_id = self.trainset.to_raw_iid(i)
item_features = self.content_features.loc[raw_item_id:raw_item_id, :].values
gradient_boosting = self.user_profile[u]
score = gradient_boosting.predict(item_features)[0]
elif self.regressor_method == 'random_forest':
raw_item_id = self.trainset.to_raw_iid(i)
item_features = self.content_features.loc[raw_item_id:raw_item_id, :].values
randomforest = self.user_profile[u]
score = randomforest.predict(item_features)[0]
else :
score = None
# (implement here the regressor prediction)
return score
def explain(self, u) :
if u in self.user_profile_explain :
return self.user_profile_explain[u]
else :
return {}
cb = ContentBased("movie_year", "random_sample")
print(cb.explain('11'))
print('test')
```
%% Output
{}
%% Cell type:code id:baab88b7 tags:
``` python
from pprint import pprint
# Créer une instance de TfidfVectorizer pour les genres
tfidf_vectorizer = TfidfVectorizer()
# Fit et transform pour calculer la matrice TF-IDF des genres
tfidf_matrix = tfidf_vectorizer.fit_transform(df_items['genres'])
# Obtenir les noms des genres (features)
genre_names = tfidf_vectorizer.get_feature_names_out()
# Créer un DataFrame à partir de la matrice TF-IDF des genres
df_tfidf = pd.DataFrame(tfidf_matrix.toarray(), columns=genre_names)
print("Matrice TF-IDF des genres :")
display(df_tfidf)
```
%% Output
Matrice TF-IDF des genres :
%% Cell type:markdown id:ffd75b7e tags:
The following script test the ContentBased class
%% Cell type:code id:69d12f7d tags:
``` python
def test_contentbased_class(feature_method, regressor_method):
"""Test the ContentBased class.
Tries to make a prediction on the first (user,item ) tuple of the anti_test_set
"""
sp_ratings = load_ratings(surprise_format=True)
train_set = sp_ratings.build_full_trainset()
content_algo = ContentBased(feature_method, regressor_method)
content_algo.fit(train_set)
anti_test_set_first = train_set.build_anti_testset()[0]
prediction = content_algo.predict(anti_test_set_first[0], anti_test_set_first[1])
print(prediction)
# print("title_length :")
# test_contentbased_class(feature_method = "title_length" , regressor_method = "random_score")
# test_contentbased_class(feature_method = "title_length" , regressor_method = "random_sample")
# test_contentbased_class(feature_method = "title_length" , regressor_method = "linear_regression")
# test_contentbased_class(feature_method= "title_length", regressor_method= "svr_regression")
# test_contentbased_class(feature_method= "title_length", regressor_method= "gradient_boosting")
# test_contentbased_class(feature_method= "title_length", regressor_method= "random_forest")
# print("\n")
# print("movie_year : ")
# test_contentbased_class(feature_method= "movie_year", regressor_method= "random_score")
# test_contentbased_class(feature_method= "movie_year", regressor_method= "random_sample")
# test_contentbased_class(feature_method= "movie_year", regressor_method= "linear_regression")
# test_contentbased_class(feature_method= "movie_year", regressor_method= "svr_regression")
# test_contentbased_class(feature_method= "movie_year", regressor_method= "gradient_boosting")
# test_contentbased_class(feature_method= "movie_year", regressor_method= "random_forest")
# print("\n")
# print("relevance : ")
# test_contentbased_class(feature_method= "relevance", regressor_method= "random_score")
# test_contentbased_class(feature_method= "relevance", regressor_method= "random_sample")
# test_contentbased_class(feature_method= "relevance", regressor_method= "linear_regression")
# test_contentbased_class(feature_method= "relevance", regressor_method= "svr_regression")
# test_contentbased_class(feature_method= "relevance", regressor_method= "gradient_boosting")
# test_contentbased_class(feature_method= "relevance", regressor_method= "random_forest")
# print("\n")
# print("genres : ")
# test_contentbased_class(feature_method= "genres", regressor_method= "random_score")
# test_contentbased_class(feature_method= "genres", regressor_method= "random_sample")
# test_contentbased_class(feature_method= "genres", regressor_method= "linear_regression")
# test_contentbased_class(feature_method= "genres", regressor_method= "svr_regression")
# test_contentbased_class(feature_method= "genres", regressor_method= "gradient_boosting")
# test_contentbased_class(feature_method= "genres", regressor_method= "random_forest")
# print("\n")
# print("rating : ")
# test_contentbased_class(feature_method= "rating", regressor_method="random_score")
# test_contentbased_class(feature_method= "rating", regressor_method="random_sample")
# # test_contentbased_class(feature_method= "rating", regressor_method="linear_regression")
# #test_contentbased_class(feature_method="rating", regressor_method="svr_regression")
# #test_contentbased_class(feature_method="rating", regressor_method="gradient_boosting")
# #test_contentbased_class(feature_method="rating", regressor_method="random_forest")
# print("\n")
# print("tags : ")
# test_contentbased_class(feature_method="tags", regressor_method="random_score")
# test_contentbased_class(feature_method="tags", regressor_method="random_sample")
# #test_contentbased_class(feature_method="tags", regressor_method="linear_regression")
# # test_contentbased_class(feature_method="tags", regressor_method="svr_regression")
# # test_contentbased_class(feature_method="tags", regressor_method="gradient_boosting")
# # test_contentbased_class(feature_method="tags", regressor_method="random_forest")
# print("\n")
# print("tags_length : ")
# test_contentbased_class(feature_method="tags_length", regressor_method="random_score")
# test_contentbased_class(feature_method="tags_length", regressor_method="random_sample")
# test_contentbased_class(feature_method="tags_length", regressor_method="linear_regression")
# test_contentbased_class(feature_method="tags_length", regressor_method="svr_regression")
# test_contentbased_class(feature_method="tags_length", regressor_method="gradient_boosting")
# test_contentbased_class(feature_method="tags_length", regressor_method="random_forest")
# print("\n")
# print("combination : ")
# test_contentbased_class(feature_method="combination", regressor_method="random_score")
# test_contentbased_class(feature_method="combination", regressor_method="random_sample")
# test_contentbased_class(feature_method="combination", regressor_method="linear_regression")
# test_contentbased_class(feature_method="combination", regressor_method="svr_regression")
# test_contentbased_class(feature_method="combination", regressor_method="gradient_boosting")
# test_contentbased_class(feature_method="combination", regressor_method="random_forest")
```
......
0% ou .
You are about to add 0 people to the discussion. Proceed with caution.
Terminez d'abord l'édition de ce message.
Veuillez vous inscrire ou vous pour commenter