Comparison of Pedestrian Count Expansion Methods: Land Use Groups versus Empirical Clusters
Comparison of Pedestrian Count Expansion Methods: Land Use Groups versus Empirical Clusters
复制标题
行人计数扩展方法的比较：土地利用组与经验集群
DOI：
10.1177/0361198118793006
复制
发表时间：
2018
影响因子：
1.7
通讯作者：
Offer Grembek
中科院分区：
文献类型：
--
作者：
Julia B. Griswold;Aditya Medury;R. Schneider;Offer Grembek
基于长期计数数据趋势的扩展系数是从短期计数中估计每日、每周或年度数量的有用工具，但不清楚如何按活动模式区分地点。本文比较了两种方法来开发因素组的小时到周的行人计数扩展因素。土地利用（LU）分类方法假设周围的LU会影响某个位置的行人活动，并且很容易根据该地点的可识别属性应用于短期计数位置。经验聚类（EC）方法使用统计方法来匹配基于实际计数的位置，这可能会产生更准确的数量估计，但提出了一个挑战，确定哪个因素组适用于一个位置。我们发现，LU和EC的方法提供了更好的每周行人流量估计比单因素的方法，采取所有位置的平均值。此外，LU和EC估计误差之间的差异是适度的，因此使用直观和实用的LU方法可能是有益的。LU分组也可以根据EC结果的见解进行修改，从而在提高估计值的同时保持应用的易用性。短期计数的理想时间是在活动高峰期，因为它们通常产生比非高峰期误差更小的估计。根据较长持续时间的计数估计的每周数量（例如，12小时）通常比来自较短持续时间计数的估计更准确（例如，2 h）。从业人员可以遵循此指南，以提高每周行人流量估计的质量。
Expansion factors based on the trends in long-term count data are useful tools for estimating daily, weekly, or annual volumes from short-term counts, but it is unclear how to differentiate locations by activity pattern. This paper compares two approaches to developing factor groups for hour-to-week pedestrian count expansion factors. The land use (LU) classification approach assumes that surrounding LUs affect the pedestrian activity at a location, and it is easy to apply to short-term count locations based on identifiable attributes of the site. The empirical clustering (EC) approach uses statistical methods to match locations based on the actual counts, which may produce more accurate volume estimates, but presents a challenge for determining which factor group to apply to a location. We found that both the LU and EC approaches provided better weekly pedestrian volume estimates than the single factor approach of taking the average of all locations. Further, the differences between LU and EC estimation errors were modest, so it may be beneficial to use the intuitive and practical LU approach. LU groupings can also be modified with insights from the EC results, thus improving estimates while maintaining the ease of application. Ideal times for short-term counts are during peak activity periods, as they generally produce estimates with fewer errors than off-peak periods. Weekly volume estimated from longer-duration counts (e.g., 12 h) is generally more accurate than estimates from shorter-duration counts (e.g., 2 h). Practitioners can follow this guidance to improve the quality of weekly pedestrian volume estimates.