Aprende Collaborative Filtering with ALS

Desliza para mostrar el menú

Alternating Least Squares (ALS) factorizes the user-item ratings matrix into two smaller matrices – one for users and one for items. Each user and item is represented as a dense vector of latent factors. The dot product of a user vector and an item vector predicts the rating.

Setting Up a Ratings Dataset

The flights dataset does not have ratings. For this chapter you will use a synthetic ratings dataset to demonstrate ALS, then apply it to the MovieLens dataset in the capstone:


              1234567891011121314151617181920212223242526
            
from pyspark.sql import SparkSession
from pyspark.ml.recommendation import ALS
from pyspark.sql import Row

spark = SparkSession.builder \
    .appName("ALS") \
    .master("local[*]") \
    .getOrCreate()

# Synthetic user-movie ratings
ratings_data = [
    Row(userId=1, movieId=1, rating=5.0),
    Row(userId=1, movieId=2, rating=3.0),
    Row(userId=1, movieId=3, rating=4.0),
    Row(userId=2, movieId=1, rating=4.0),
    Row(userId=2, movieId=4, rating=5.0),
    Row(userId=3, movieId=2, rating=4.0),
    Row(userId=3, movieId=3, rating=5.0),
    Row(userId=3, movieId=4, rating=3.0),
    Row(userId=4, movieId=1, rating=3.0),
    Row(userId=4, movieId=3, rating=4.0),
    Row(userId=4, movieId=4, rating=5.0),
]

ratings_df = spark.createDataFrame(ratings_data)
train_df, test_df = ratings_df.randomSplit([0.8, 0.2], seed=42)

Training ALS


              12345678910111213
            
als = ALS(
    userCol="userId",
    itemCol="movieId",
    ratingCol="rating",
    rank=10,            # Number of latent factors
    maxIter=10,
    regParam=0.1,
    coldStartStrategy="drop"  # Dropping users/items not seen during training
)

model = als.fit(train_df)
predictions = model.transform(test_df)
predictions.show()

coldStartStrategy="drop" removes rows with users or items not present in the training set – without this, predictions for unseen entities are NaN.

Generating Top-N Recommendations


              1234567
            
# Top 3 movie recommendations per user
user_recs = model.recommendForAllUsers(3)
user_recs.show(truncate=False)

# Top 3 user recommendations per movie
movie_recs = model.recommendForAllItems(3)
movie_recs.show(truncate=False)

¿Todo estuvo claro?

¡Gracias por tus comentarios!

Sección 1. Capítulo 11

Pregunte a AI

Pregunte lo que quiera o pruebe una de las preguntas sugeridas para comenzar nuestra charla

Sección 1. Capítulo 11