你以为是“说走就走”？其实是“算好才走”：大数据是怎么悄悄优化旅游体验的？

import pandas as pd

# 假设我们有一个用户行为数据集
df = pd.read_csv('user_behavior.csv')

# 提取核心特征
user_profile = df.groupby('user_id').agg({
    'search_location': lambda x: x.mode()[0],
    'booking_count': 'sum',
    'avg_spend': 'mean',
    'travel_type': lambda x: x.mode()[0],  # 自由行、跟团游
})

# 打标签
def tag_user(row):
    if row['avg_spend'] > 3000:
        return '高消费型'
    elif row['travel_type'] == '自由行':
        return '个性偏好型'
    else:
        return '节约理性型'

user_profile['user_tag'] = user_profile.apply(tag_user, axis=1)

from pyspark.sql import SparkSession
from pyspark.sql.functions import col, explode, split

spark = SparkSession.builder.appName("TourismTrend").getOrCreate()
df = spark.read.json("weibo_travel_data.json")

# 分词提取热词
keywords = df.select(explode(split(col("content"), " ")).alias("keyword"))
hot_keywords = keywords.groupBy("keyword").count().orderBy("count", ascending=False)

hot_keywords.show(10)

from surprise import SVD, Dataset, Reader
from surprise.model_selection import train_test_split

# 构建评分数据（用户对景区、酒店、餐厅的评分）
reader = Reader(rating_scale=(1, 5))
data = Dataset.load_from_df(df[['user_id', 'place_id', 'rating']], reader)

trainset, testset = train_test_split(data, test_size=0.2)
model = SVD()
model.fit(trainset)

# 推荐景点
user_id = 'U123'
top_places = []
for place_id in df['place_id'].unique():
    pred = model.predict(user_id, place_id)
    top_places.append((place_id, pred.est))

top_places = sorted(top_places, key=lambda x: x[1], reverse=True)[:5]