Abstract：
Hot spot analysis of linked accelerometer and Global Positioning System data is often used to identify areas of high/low activity in the schoolyard. We illustrate the potential impact of a suite of methodological decisions (i) accelerometer metric; (ii) monitor epoch; (iii) number of recess periods/days and level of aggregation; (iv) sample size; (v) distance band; (vi) spatial versus spatiotemporal weighting scheme; and (vii) time band. Accelerometer metrics resulted in different clustering patterns. Longer epochs resulted in a less detailed picture of schoolyard behavior. Level of data aggregation impacted cluster patterns due to inter-period and inter-day differences, but clusters were consistent with increasing sample size. Use of spatiotemporal weight matrices resulted in better separation of hot and cold spots and revealed potentially important temporal clustering patterns. Increasing distance or time band resulted in reallocation of small clusters to larger clusters. Hot spot analysis decisions should be clearly reported in future studies.
摘要：
链接的加速度计和全球定位系统数据的热点分析通常用于识别校园中的高/低活动区域。我们说明了一系列方法决策的潜在影响（i）加速度计度量；（ii）监测时期；（iii）休会期/天数和聚集水平；（iv）样本量；（v）距离带；（vi）空间与时空加权方案；（vii）时间带。加速度计度量导致不同的聚类模式。更长的时期导致对校园行为的详细了解。由于跨期和日之间的差异，数据聚合水平影响了集群模式，但是聚类与样本量的增加是一致的。使用时空权重矩阵可以更好地分离热点和冷点，并揭示出潜在的重要时间聚类模式。增加距离或时间带导致将小簇重新分配给较大簇。热点分析决策应在未来的研究中明确报告。
