Abstract: The parking demand forecasting model at residential was studied by two steps method, which was the combination of multivariate regression method and correction coefficient method. Two steps method made up for the deficiency of traditional multiple linear regression and elastic coefficient method. Based on the residential parking survey, the parking demand forecasting model was established, which combined the multivariate regression prediction, location correction coefficient with building type correction coefficient. The predictive values were taken as the basis of a new similar residential parking resource. Two steps method not only fully considers the influence of various factors on the parking demand forecast at residential, but also has a flexible control for the location and building types, which has certain practicability and flexibility in practice. Taking Tangshan city for example, the residential areas were selected to validate the model, and the errors were all less than 10%, which verified the effectiveness of the forecast model.

Key words: traffic and transportation engineering, residential area, parking demand, parking, two steps method

摘要： 研究居住区停车泊位需求预测模型，采用多元回归与修正系数法相结合的二步骤法，改变传统的多元线性回归和弹性系数法的不足。提出以居住区停车调查为基础，建立停车泊位需求预测模型，将多元回归预测和区位修正系数、建筑类型修正系数相结合，以此预测值作为新建同类居住区停车资源配备的依据。二步骤法既充分考虑了各种影响因素对居住区泊位需求的影响，又对区位和建筑物类型等因子具有一定的弹性控制，实践中有一定的实用性和灵活性。以唐山市为例，选取居住区进行模型方法验证，误差均﹤10%，说明模型预测的有效性。

关键词: 交通运输工程, 居住区, 泊位需求, 停车, 二步骤法