Introduction

Increasing urban population and traffic density on one side and rising hospital demands on the other has underlined the importance of transportation modeling. In this study, Four-Step Transportation Model (FSTM), which focuses primarily on home-based work and school trips, is used to evaluate the increasing share of home-based hospital trips among all from a transportation planning perspective. The aims of this study are to 1) examine the hospital trip behaviors and the parameters affecting it within the framework of ‘home-based hospital trips’, 2) evaluate the effectiveness of different robust and biased estimation techniques to be used which could be an alternative to Ordinary Least Square (OLS) in FSTM.

Method

OLS, Ridge Regression (RR), Least Trimmed Squares (LTS), and Least Trimmed Squares-Ridge (LTS-Ridge) techniques were used for the comprehensive evaluation of home-based hospital trip production models for the Eskişehir City. In this context, the five characteristics affecting hospital trips were used as independent variables. Approximately 20000 valid household survey data (HSD) for 2001 (Training data) and 29000 valid HSD for 2015 (Testing data) were used and results were evaluated in terms of Mean Squared Error (MSE).

Results

As a result of the analysis, the MSE values of LTS-Ridge, LTS, RR, and OLS models are 127484, 169060, 274211, 434164, respectively. The most consistent and successful results were obtained from LTS-Ridge according to MSE and direction of the coefficients. Hospital demand coefficient proposed in this study increased the success of future estimations.

Conclusion

When data have multicollinearity or contain outliers, LTS-Ridge makes more successful predictions than OLS. This study fills a large gap in the literature by examining the home-based hospital trips in terms of socio-economic and demographic characteristics from a transportation planning perspective.

中文翻译：

---

根据交通规划估算出行特征

介绍

一方面，城市人口和交通密度的增加，另一方面，医院需求的增加，突显了交通模型的重要性。在这项研究中，主要针对家庭工作和学校旅行的四步运输模型（FSTM），用于从运输计划的角度评估所有家庭中医院旅行的增加份额。这项研究的目的是：1）在“家庭式医院出诊”框架内检查医院出诊行为和影响其的参数，2）评估要使用的各种可靠且有偏见的估计技术的有效性，这可能是一种替代FSTM中的普通最小二乘（OLS）。

方法

使用OLS，岭回归（RR），最小二乘平方（LTS）和最小二乘平方-里奇（LTS-Ridge）技术对埃斯基谢希尔市的家庭医院出行生产模型进行了综合评估。在这种情况下，影响医院出行的五个特征被用作自变量。大约使用了2001年（培训数据）的20000个有效家庭调查数据（HSD）和2015年（测试数据）的29000个有效HSD（测试数据），并根据均方误差（MSE）评估了结果。

结果

分析的结果是，LTS-Ridge，LTS，RR和OLS模型的MSE值分别为127484、169060、274211和434164。根据MSE和系数的方向，从LTS-Ridge获得了最一致和成功的结果。本研究提出的医院需求系数增加了未来估计的成功率。

结论

当数据具有多重共线性或包含异常值时，LTS-Ridge会比OLS进行更成功的预测。这项研究通过从交通规划的角度从社会经济和人口特征的角度考察了家庭医院的出行，填补了文献中的巨大空白。