微博签到数据的时空热点挖掘——以北京为例

摘要

对社交媒体位置服务大数据进行时空数据挖掘能为城市规划、商业决策、用户行为分析等应用提供决策依据.基于新浪微博签到点数据,应用Knox指数进行时空交互性检验,确定合适的时空分析尺度,并利用时空重排扫描统计方法分别在短时间尺度(时)和长时间尺度(天)下挖掘时空热点.结果表明:短时间尺度(时)和长时间尺度(天)的签到点都具有随着空间距离的增大,时空交互性逐渐增强的趋势;短时间尺度下的时空热点区域主要分布在主城区,覆盖半径集中在2~6 km、时间集中在11:00—17:00,热点持续时长约为3~5 h;长时间尺度下的时空热点主要集中在主城区,少量均匀分布在城外,覆盖半径集中在5~6 km,时间集中在2016-02-07—2016-02-13,热点持续时长约为3~6d.%Mining spatial-temporal hot spots from check-in points can provide decision-making reference for urban planning and management,business strategy,user behavior analysis and other applications.This paper uses Sina microblog open platform to obtain check-in data,with Beijing as an example.By choosing the extended Knox index for space-time interaction test,the appropriate analytical scale then can be derived.Next,the space-time permutation scan statistic model is used to mine spatial-temporal hot spots under short time scale(hours)and long time scale(days).The study indicates that:a.In the short time scale(hour)and the long time scale(day),with the increase of the spatial scale,the space-time interactivity gradually increases;b.From short time scale,the spatial-temporal hot spots are mainly distributed in the central urban area with a coverage about 2~6 km,occurring time is concentrated on 11 a.m. ~17 p.m.,and the duration-time is about 3~5 hours;c.From long time scale,the spatial-temporal hot spots are mainly distributed in the central urban area with a coverage about 5~6 km,occurring time is concentrated on 2016/2/7-2016/2/13,and the duration-time is about 3-6 days.