中原地区多日不用气及户内安全风险告警数据

随着社会逐步进入老龄化社会，面临的独居老人生活问题越来越受到政府、爱心人士的关心。燃气公司通过户主年龄，户内用气量与平均用气量比较，推断用户是否为独居老人。结合日用气量的连续特征，判断燃气用户用气是否正常，进一步及时发现户内老人发生安全风险的情况。1、通过获取要素字段数据：用气时间、日用气量、户号、地址区域、地址区域内用户数、平均每户正常用气量、近期缴费方式（柜台、线上缴费）、开户时间、通气时间等数据
2、数据清洗，处理缺失值和异常值，并标注是否为独居老人（60岁以上且用气量显著低于区域平均值）。
3. 风险预警模型构建
使用ARIMA模型对日用气量数据进行时序分析，识别独居老人。结合时序分析和深度学习预测结果，建立综合的安全风险评估模型。采用K-means聚类算法，根据用户的地址区域、用户数、平均每户正常用气量等特征，识别异常用气行为。

As society gradually enters an aging era, the living issues of elderly people living alone have attracted increasing attention from governments and caring individuals. Gas utilities infer whether a user is an elderly person living alone by comparing the householder's age and indoor gas consumption with the regional average gas consumption. Combined with the continuous features of daily gas consumption, they judge whether the gas usage of the user is normal, so as to timely detect potential safety risks faced by the elderly in the household.

1. Feature field data collection: Collect data including gas usage time, daily gas consumption, household ID, address region, number of users within the address region, normal average gas consumption per household, recent payment methods (counter, online payment), account opening time, gas connection time, etc.

2. Data cleaning: Process missing values and outliers, and label whether the user is an elderly person living alone (aged 60 or above with gas consumption significantly lower than the regional average).

3. Risk early warning model construction: Conduct time-series analysis on daily gas consumption data using the ARIMA model to identify elderly people living alone. Establish a comprehensive safety risk assessment model by combining time-series analysis and deep learning prediction results. Adopt the K-means clustering algorithm to identify abnormal gas usage behaviors based on features such as the user's address region, number of users, normal average gas consumption per household, etc.

搜集汇总

数据集介绍

特点

该数据集记录了中原地区燃气用户的用气数据及安全风险告警信息，主要用于识别独居老人并预警潜在的安全风险。数据规模为21126条，每日更新，应用场景包括通过用气数据分析用户行为，及时发现异常情况。