双十一移动推送推荐

import pandas as pd
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.metrics.pairwise import linear_kernel

# 假设我们有一个用户行为数据集
data = {
    'user_id': [1, 1, 2, 2],
    'product_id': [101, 102, 103, 104],
    'action': ['view', 'purchase', 'view', 'view']
}
df = pd.DataFrame(data)

# 产品描述数据
products = {
    101: 'Smartphone with advanced camera',
    102: 'Laptop with high performance processor',
    103: 'Tablet with long battery life',
    104: 'Smartwatch with fitness tracking'
}

# 将产品描述转换为TF-IDF向量
tfidf = TfidfVectorizer(stop_words='english')
tfidf_matrix = tfidf.fit_transform(products.values())

def get_recommendations(user_id):
    user_actions = df[df['user_id'] == user_id]
    viewed_products = user_actions[user_actions['action'] == 'view']['product_id'].tolist()
    purchased_products = user_actions[user_actions['action'] == 'purchase']['product_id'].tolist()
    
    if purchased_products:
        product_index = purchased_products[0] - 100
        cosine_similarities = linear_kernel(tfidf_matrix[product_index:product_index+1], tfidf_matrix).flatten()
        related_products_indices = cosine_similarities.argsort()[:-5:-1]
        recommendations = [list(products.keys())[i] for i in related_products_indices if list(products.keys())[i] not in viewed_products]
        return recommendations
    else:
        return []

# 示例：为用户1推荐产品
recommendations = get_recommendations(1)
print("Recommended products for user 1:", recommendations)