Files
airflow/dags/stats/community/rate.py
T
2026-07-22 15:17:02 +08:00

53 lines
2.8 KiB
Python

"""
小区周边1公里交通地铁入库业务逻辑
"""
import datetime
import logging
import pandas as pd
from goojia_utils import MAIN_MONGO_CLIENT
from goojia_utils import dateutils
from sklearn.mixture import GaussianMixture
from pymongo import InsertOne, DeleteMany, ReplaceOne, UpdateOne
logger = logging.getLogger(__name__)
district_stat_conn = MAIN_MONGO_CLIENT.get_collection('fangjia_stat', 'district_stat')
district_analy_data_conn = MAIN_MONGO_CLIENT.get_collection('fangjia_stat', 'district_analy_data')
server_area_conn = MAIN_MONGO_CLIENT.get_collection('fangjia', 'server_area')
weekend = list(district_stat_conn.find().sort([("weekend", -1)]).limit(1))[0]['weekend']
level_map = {"1": "S", "2": "A", "3": "B", "4": "C", "5": "D"}
logger.info(f"开始计算楼盘评级 weekend:{dateutils.format_date(weekend, '%Y-%m-%d')}")
for server_area in server_area_conn.find({'display': {'$ne': '全国'}}, {"display": 1}):
city = server_area["display"]
data = pd.DataFrame(
district_stat_conn.find({"city": city, "weekend": weekend},
{"_id": 0, "region": 1, "district_name": 1, "avg_price": 1}))
if "avg_price" not in data or len(data['avg_price'].unique()) < 5:
continue
data['city'] = city
logger.info(f"开始计算{city}楼盘评级 weekend:{dateutils.format_date(weekend, '%Y-%m-%d')}")
# 高斯混合模型
gmm = GaussianMixture(n_components=5, covariance_type='tied', max_iter=200, n_init=10).fit(
data[["avg_price"]])
data.loc[:, 'cluster'] = gmm.predict(data[["avg_price"]])
data.loc[:, 'price_mean_cluster'] = data.groupby('cluster')["avg_price"].transform('mean')
data.loc[:, "target"] = data['price_mean_cluster'].rank(method='dense', ascending=False).astype(int)
data = data.drop(columns=['cluster', 'price_mean_cluster'], axis=1)
data["community_rating"] = data['target'].apply(lambda x: level_map[str(x)])
opts = []
for item in data.to_dict('records'):
opts.append(UpdateOne({"city": city, "region": item["region"], "district_name": item["district_name"]},
{"$set": {"community_rating": item["community_rating"]}}, upsert=True
))
# 按照每 1000 条分割 updates 列表
batch_size = 1000
batches = [opts[i:i + batch_size] for i in range(0, len(opts), batch_size)]
# 逐个批次执行批量更新操作
cnt = 0
for batch in batches:
cnt = cnt + 1
logger.info(f"{city}楼盘评级更新数据库, 进度:{cnt}/{len(batches)}")
result = district_analy_data_conn.bulk_write(batch)
print(f"当前批次更新的文档数量: {result.modified_count}")
logger.info(f"楼盘评级计算结束{dateutils.format_date(datetime.datetime.utcnow(), '%Y-%m-%d')}")