Weather factor dummy variable.

This paper studies the parking demand characteristics of large commercial areas in the city’s central regions. The study uses non-parametric and semi-parametric analysis methods in survival analysis to explore if and how weather conditions, parking tariffs, and temporal factors (weekdays, weekends, and short holidays) impact the parking duration. The parking data of a large commercial supermarket in Zhengzhou was collected over one month. Single-factor analysis based on the Product-Limit (PL) approach suggests that the cumulative survival and relative risk curves of parking duration exhibit slight variations across different temporal categories and weather conditions. Based on Cox semi-parametric multi-factor analysis results, the parking duration is significantly influenced by weekdays (regression coefficient = 0.068, hazard ratio = 1.071, P < 0.001), weekends (regression coefficient = 0.042, hazard ratio = 1.043,P < 0.001), moderate rain (regression coefficient = -0.089, hazard ratio = 0.914, P < 0.001), and heavy rain (regression coefficient = 0.030, hazard ratio = 1.030,P = 0.034 < 0.05). The results have indicated that within the study area, compared to short holidays, the parking duration on weekdays and weekends is shorter, with the probability of vehicles ending their parking increased by 7.1% and 4.3%, respectively. Under different weather conditions, compared to sunny days, parking duration is longer during moderate rain, with the probability of vehicles departing decreased by 8.6%, whereas during heavy rain, parking duration is shorter, with the probability of vehicles departing increased by 3%. Notably, the parking tariffs demonstrated no statistically significant impact. These findings suggest that temporal variations and rainfall patterns should inform dynamic parking management strategies, including weather-responsive pricing adjustments and spatial capacity optimization during peak periods.本研究聚焦城市中心城区大型商业区域的停车需求特征。本研究采用生存分析（survival analysis）中的非参数与半参数分析方法，探究天气状况、停车费率以及时间因素（工作日、周末及短假期）是否以及如何对停车时长产生影响。研究采集了郑州某大型商业超市为期一个月的停车数据。基于乘积限（Product-Limit, PL）法的单因素分析结果显示，不同时间类别与天气条件下，停车时长的累积生存曲线与相对风险曲线仅存在轻微差异。基于Cox半参数多因素分析（Cox semi-parametric multi-factor analysis）结果，停车时长显著受以下因素影响：工作日（回归系数=0.068，风险比=1.071，P<0.001）、周末（回归系数=0.042，风险比=1.043，P<0.001）、中雨（回归系数=-0.089，风险比=0.914，P<0.001）以及大雨（回归系数=0.030，风险比=1.030，P=0.034<0.05）。研究结果表明，在本次研究区域内，相较于短假期，工作日与周末的停车时长更短，车辆结束停车的概率分别提升7.1%与4.3%。在不同天气条件下，相较于晴天，中雨天气下停车时长更长，车辆驶离的概率降低8.6%；而大雨天气下停车时长更短，车辆驶离的概率提升3%。值得注意的是，停车费率未表现出统计显著性影响。上述研究结果提示，时间维度的变化与降雨模式可为动态停车管理策略提供依据，包括响应天气的费率调整以及高峰时段的空间容量优化。