独居老人用电行为分析数据

通过电力大数据分析，了解独居老人的用电习惯和规律，识别异常用电行为，及时发出风险预警，有效提升对独居老人群体精细化管理和服务水平，助力社区关怀、子女远程照护。1、数据清洗：包括异常数据处理（去除分时功率中的极大值）和日电量缺失填充（结合分时功率进行填充）。
2、数据转换：将电力用户和民政的独居老人进行关联并生成唯一标识，即独居老人和预警id。
预警原因：运行决策树算法将用电数据转换为预警原因：①电量数据缺失：当天没有采集到用电数据；②日电量为0：当天采集到用电量为0；③初筛居家：日电量＞历史日电量中位数且日电量高于3度、当天存在至少两个用电高峰（＞0.5kw）等四个维度；④分时功率平稳：当天用户日电量＞历史在家最低日电量的80%、当天功率多样性较低等三个维度；⑤在家：除以上情况。
3、数据加工：
将当天时间段分割为0点-6点、6点-12点、12点-18点、18点-24点四个。
每个时间段平均功率=时间段功率和/有效采集点数
时间段平均功率占比=时间段平均功率/当日各时间段平均功率之和
居家状态：由系统决策树判定算法综合判断：
预警原因【初筛居家】、【在家】：值为0，表示老人正常用电；
预警原因【分时功率平稳】、【日电量为0】：值为1，表示老人没有用电或者没有使用过需要主动启动的电器，需要特别关注；
预警原因【电量数据缺失】：值为2，数据无效

By conducting big data analysis on electricity consumption data, this dataset aims to uncover the electricity usage habits and patterns of elderly individuals living alone, identify abnormal electricity consumption behaviors, issue timely risk warnings, effectively enhance the refined management and service capabilities for the elderly living alone, and facilitate community care and remote care provided by their children.
1. Data Cleaning: This includes two sub-steps: abnormal data processing (removing extreme values from time-sharing power data) and missing daily electricity quantity imputation (imputing missing values based on time-sharing power data).
2. Data Transformation: Associate electricity consumption users with civil affairs records of elderly people living alone, and generate unique identifiers, specifically the unique IDs for elderly individuals living alone and warning records.
Warning Reason Generation: Convert electricity consumption data into warning reasons using a decision tree algorithm, covering the following five scenarios:
① Missing Electricity Data: No electricity consumption data was collected on the target day;
② Zero Daily Electricity Consumption: The collected electricity consumption volume is 0 on the target day;
③ Preliminary Identified as Staying at Home: Daily electricity consumption exceeds the median of historical daily electricity consumption, daily electricity consumption is higher than 3 kWh, and there are at least two electricity consumption peaks (>0.5 kW) on the target day, etc. (four dimensions in total);
④ Stable Time-sharing Power Consumption: The daily electricity consumption on the target day exceeds 80% of the historical minimum daily electricity consumption when the elderly are staying at home, and the power diversity on the day is relatively low, etc. (three dimensions in total);
⑤ Staying at Home: All other cases not covered by the above four scenarios.
3. Data Processing:
Split the daily time period into four segments: 0:00–6:00, 6:00–12:00, 12:00–18:00, and 18:00–24:00.
- Average power per time segment = sum of power values within the time segment / number of valid collection points
- Proportion of average power per time segment = average power of the time segment / sum of average power across all four time segments on the target day
Home Status Judgment: Conducted comprehensively via the systematic decision tree algorithm:
- For warning reasons [Preliminary Identified as Staying at Home] and [Staying at Home]: The home status value is 0, indicating normal electricity consumption by the elderly;
- For warning reasons [Stable Time-sharing Power Consumption] and [Zero Daily Electricity Consumption]: The home status value is 1, indicating that the elderly either did not use electricity or did not use actively activated appliances, requiring special attention;
- For warning reason [Missing Electricity Data]: The home status value is 2, indicating invalid data.

搜集汇总

数据集介绍

特点

该数据集由国网浙江省电力有限公司杭州供电公司提供，包含500条独居老人的用电行为数据，每日更新。数据通过电力大数据分析独居老人的用电习惯，识别异常行为并提供预警，旨在提升对独居老人群体的精细化管理和服务水平。