论文 Papers

基于消费者行为的商业空间绩效评价体系建构. 王灿, 王德, 朱玮, 宋姗. 城市规划, 2021, (3): 9-21.
Construction of assessment system on commercial space performance based on consumer behavior. Can Wang, De Wang, Wei Zhu, Shan Song. City Planning Review, 2021, (3): 9-21.
►  链接 Hyperlink  |  摘要 Abstract 商业空间绩效需要科学的评价体系，而既有研究对消费者行为的关注不足。在分析商业空间各方利益的基础上，从4个维度构建了消费者行为视角下的评价体系：个体行为链指标评价了总体活力与效率；空间分区指标评价了各局部空间自身的活力与效率；到访特定空间的特定消费者的行为链指标评价了各局部空间对于整体活力的贡献；结构性指标评价了商业空间各组成要素的相关关系，包括空间流动性、均好性，功能网络的带动效应。利用该体系，研究对上海市五角场万达广场开展评价，认为其总体情况较理想，但是也存在若干问题，包括垂直空间结构较封闭、部分空间利用不充分、日用品功能的面积偏高、活动分布不均衡等，并提出了相应的空间优化建议。
A scientific assessment system is needed for commercial space performance, while too limited attention was paid to consumer behavior in existed studies. This paper analyzes the interests of different parties relating to commercial space, and constructs a consumer-behavior-based assessment system from 4 dimensions. Firstly, the indices of individual activity chains assess the overall vibrancy and efficiency of the whole space. Secondly, the indices of spatial zones focus on vibrancy and efficiency of each local space, which is limited in itself. The third kind of indices are concerning individual activity chains of specific consumers who visit certain local space, which focus on contributions that local space makes to the whole space. Finally, the structural indices assess the relationship between different components, including flowability and equality among spaces, and mutual support in the network of multiple functions. The advised assessment system is utilized in the empirical study of Shanghai WJC Wanda Square. The results suggest that its spatial performance is satisfactory in general, but some problems exist, including isolation in vertical space system, underutilization of some local spaces, excessive area of business of daily supplies, imbalanced spatial distribution of activities, and so on. Corresponding measures for spatial improvement are suggested.

Understanding consumers’ spatial behavior in urban commercial environment by extracting typical routes. Can Wang, De Wang, Shan Song, Wei Zhu. ACSP (Association of Collegiate Schools of Planning) Annual Conference, Toronto, Canada, November, 2020.
►  链接 Hyperlink  |  摘要 Abstract This paper proposed two approaches to extract typical pedestrian routes from consumers’ multi-stop spatial behavior, which is quite complex but undoubtedly essential for urban design of commercial space. The first approach is to conduct a luster analysis for individual routes. The degree of similarity of each pair of routes is measured by Levenstein ratio, and affinity propagation algorithm is applied to the similarity matrix to generate exemplars as typical routes, from which different patterns of spatial behavior could be revealed. The second approach is to simulate a route with the maximum probability. For this purpose, two models, namely convolutional neural network (CNN) and discrete choice model (DCM), are adopted to predict the probability that a consumer will choose a specific place as the next destination. Based on that, a typical route is simulated step by step using beam search. These two approaches are applied on a dataset consist of 323 individual routes observed in a commercial complex in Shanghai, China. For the first approach, six typical routes are derived to represent different patterns, such as task shopping, fashion-oriented shopping, and so on. For the second approach, three sets of explanatory variables, namely functional attractiveness, spatial impedance, and leaving indicators, are used for both CNN and DCM. It is found that CNN performs significantly better than classical CDM in terms of cross-validation accuracy, but it is much more prone to overfitting and could hardly provide intuitive understandings of spatial behavior mechanisms. By contrast, the results of DCM are easy to interpret and provide satisfactory accuracy. The typical routes simulated by beam search suggests that consumers are mainly focusing on anchor stores, while smaller stores will also be patronized in the link between anchor stores.

Study on the spatial configuration of nursing homes for the elderly people in Shanghai: Based on their choice preference. Shan Song, De Wang, Wei Zhu, Can Wang. Technological Forecasting and Social Change, 2020, 152(3); doi: 10.1016/j.techfore.2019.119859.
► 链接 Hyperlink | 摘要 Abstract The rapid population aging arouses the enormous demands of elderly people for institutional care services. Although the supply of nursing homes is still short in quantity, a structural imbalance on the utilization of nursing homes has appeared in many cities of China recently. Namely, when elderly people wait a long time for a bed of downtown nursing homes, many suburban ones are not fully utilized. This paper aims to reveal the elderly people’s preferences for the nursing homes, in order to clarify the reason for utilization imbalance and put forward the efficient strategies for planning practice. A stated preference survey was undertaken to collect the choice data of respondents (n=293) in Shanghai, 2013. Discrete choice models were used in data analysis to explain the mechanism of preferences. All seven attributes of nursing homes significantly affect elderly people’s preferences. Disparities on the medical condition, transportation, and distance to home between downtown and suburban nursing homes are the leading cause of utilization imbalance. By optimizing the configuration of those factors, the crowded downtown elderly people could be attracted to suburban nursing homes. Finally, the mechanism was applied to a simulation of nursing home planning.

Spacial characteristics of demand and supply of elderly care institution in Shanghai. Shan Song, Can Wang. Annual Conference of Association of American Geographers, Washington DC, USA, Apr, 2019.
► 链接 Hyperlink | 摘要 Abstract As one of the first group of areas in China that step into the aging society around the beginning of 21 century, Shanghai is experiencing a rapid deep aging process of the household population. The recent trend of elderly care development is that, on one hand, traditional home-based care pattern is challenged in the current social economic context, while on the other hand, elderly people’s demand for the institution-based care pattern is increasing. With the data of China 2010 population census and Shanghai elderly care service map, this study makes an analysis of the spatial distribution of the elderly people and nursing homes in Shanghai. The result shows mainly three points. The first point is that downtown is the most population densified area where about 60% of household elderly population crowd to live. Secondly, the aging levels in downtown and outer suburbs are higher than that in inner suburbs, thus it forms a high-low-high circle structure. Thirdly, although the nursing homes seem to be accorded with the household elderly population on the characteristics of spatial distribution, downtown nursing homes shoulder much heavier pressure of elderly care service than the suburb ones.

消费者行为视角下的商业综合体功能复合效应与机制. 王灿, 王德, 朱玮, 宋姗. 新建筑, 2018, (4): 112-116.
Multi-functional effects and mechanism of commercial complex from the perspective of consumer behavior. Can Wang, De Wang, Wei Zhu, Shan Song. New Architecture, 2018, (4): 112-116.
►  链接 Hyperlink  |  摘要 Abstract 以上海市五角场万达广场为案例，从消费者行为视角探讨了商业综合体功能复合效应的内在规律。研究以消费者对多功能的组合使用作为功能联系的媒介，自下而上地建立复合功能网络；通过社会网络分析方法测度各功能的核心度，建构“核心—边缘”结构；进而按照核心度与计划性将功能划分为主导发动、活跃辅助、定向任务、随机孤立4类，对每类功能提出了配置和布局建议。研究特别指出了大型超市作为主力店的意义有被高估的风险，以及在空间布局中提取功能之间的关联规则、将关联功能邻近布置的策略。研究为商业综合体复合功能体系的组织提供支持。
Taking Shanghai Wujiaochang Wanda Plaza as example, this paper analyzes the multi-functional effect of commercial complex from the perspective of consumer behavior. Multi-functional network is built bottom-up in the way that consumers are viewed as agents delivering connections among functions via combinational usage of them. Then social network analysis methods are utilized to calculate coreness index to represent the position of functions in the network, and construct the core-periphery structure. After that, the paper divides all functions into 4 roles according to their coreness and planning proportion, for each the strategies of configuration and spatial layout are further discussed. It is specially emphasized that the significance of large supermarket as the anchor store might be overestimated, and it is suggested to extract associated functions and arrange them nearby when making spatial layout. The findings of this paper are expected to support the planning practice.

Classification and prediction of pedestrian routes in urban commercial space. Can Wang, De Wang, Wei Zhu, Shan Song. Annual Conference of Association of American Geographers, New Orleans, USA, Apr, 2018.
► 链接 Hyperlink  |   摘要 Abstract Consumers’ walking behavior always presents great complexity, and it is well believed that better understanding and reasonable prediction of pedestrian routes could be quite beneficial for urban commercial space design. Based on a random sample of 323 routes collected from a shopping mall in Shanghai China, this study calculates the similarity between each pair of two routes using edit distance, and applies affinity propagation clustering algorithm to extract exemplar routes, from which spatial behavioral modes on individual scale are revealed. The paper then splits all routes to 1761 from-to choices, and proposes 2 kinds of models to simulate how consumers choose their next destinations. One is classical discrete choice model family, including logit model and nested logit model; the other is convolutional neural network (CNN), which combines multiple logit models together nonlinearly in its convolutional layer. For both models, 3 groups of explanatory variables are used, including attractiveness, spatial impedance, and go-home indicators. It is found that CNN performs slightly better in term of cross validation accuracy, but it is much more prone to overfitting and could hardly provide intuitive understandings of spatial behavioral mechanism. In contrast, the advantages of discrete choice models are satisfactory accuracy, immunity to overfitting, significant interpretability, and simplicity.

商业综合体的消费者空间行为特征与评价. 王德, 王灿, 朱玮, 宋姗. 建筑学报, 2017, 581(2): 27-32.
Spatial features and assessment of consumer behavior in commercial complex.  De Wang, Can Wang, Wei Zhu, Shan Song. Architectural Journal, 2017, 581(2): 27-32.
► 链接 Hyperlink  |   摘要 Abstract 基于消费者行为的视角，以上海五角场万达广场为案例对商业综合体的空间行为特征与模式进行了分析评价。研究首先描述了活动的空间分布以及各空间在活动的计划性、消费、时间上的特点；进而以个体为单位总结空间行为模式，通过聚类分析发现消费者在到访空间组合上的特征差异，通过计算路径间的相似度提取典型路径，并揭示其组织规律；最后从行为视角对空间绩效进行了评价。
From the perspective of consumer behavior, this paper analyzes and assesses the spatial behavioral features and modes of Wujiaochang Wanda Plaza, a typical case of commercial complex. The paper depicts the spatial distribution of activities and related attributes, such as plan, cost and duration. It reveals spatial behavioral modes on individual scale to identify organizing features through the study of both different combinations of spaces visited using clustering methods and typical routes based on a calculation of similarities among individual routes. The paper concludes with an assessment of spatial performance from the behavioral viewpoint.

基于需求偏好的上海市养老机构空间配置研究. 宋姗, 王德, 朱玮, 王灿. 城市规划, 2016, 40(8): 77-90.
Spatial allocation study on Shanghai’s nursing facilities for the elderly: based on the elderly’s choice behavior. Shan Song, De Wang, Wei Zhu, Can Wang. City Planning Review, 40(8): 77-90.
► 链接 Hyperlink  |   摘要 Abstract 人口快速老龄化使得养老形势日益严峻，机构养老需求增大。在养老资源如此紧张的当前，上海养老机构却出现了市区“一床难求”、郊区“空置率高”的对立。研究针对这一现象，提出了以老年人需求为导向的研究思路，通过叙述性偏好调查，总结了老年人的养老机构需求特征及偏好机制，揭示了造成养老资源利用冷热不均的原因，并提出了郊区养老机构优化的策略及路径，最后以浦东新区近郊新建一处养老机构为例，开展了规划配置实践。
With the rapid process of population aging, China’s geriatric support system is facing a more serious situation. Elderly people’s demands of the nursing facilities are increasing. Although the current nursing resources are still not enough, two contradictory phenomenon have arisen in Shanghai. Namely, endowment beds are severely short in downtown, while there are still many unused beds in suburbs. In response to this problem, the paper puts forward the elderly-demand-oriented research path. Based on the stated preference survey in Shanghai, the paper has analyzed the characteristics and mechanism of the elderly’s demand preference of the nursing facilities. It reveals the reason that has led to the unbalanced demand of the endowment beds between downtown and suburbs and concludes the improving strategies and paths of suburb nursing facilities. At last, this paper introduces practice of allocation planning of nursing facilities in Chuansha Newtown, which belongs to the inner suburbs of Pudong District.

离散选择模型研究进展. 王灿, 王德, 朱玮, 宋姗. 地理科学进展, 2015, 34(10): 1275-1287.
Research progress of discrete choice models. Can Wang, De Wang, Wei Zhu, Shan Song. Progress in Geography, 2015, 34(10): 1275-1287.
► 链接 Hyperlink  |   摘要 Abstract 本文从离散选择模型(discrete choice model,DCM)体系的一般原理和应用价值出发,总结了各经典模型的基本理论和典型应用,并概括了近来年一些重要的研究新动向。多项Logit模型(multinomial logit model,MNL)是离散选择模型体系的基础,具有简洁、可靠、易实现等优点,但也存在固有的理论缺陷,由此产生了对更加精细化模型的需求。替代的精细化模型中,嵌套Logit模型(nested logit model,NL)常用于处理备选项相关、”都不选”备选项、数据合并等问题,一般极值模型(generalized extreme value model,GEV)体系是其更一般的形式;混合Logit模型(mixed logit model,MXL)可用于解决随机偏好问题和多种相关问题,包括备选项相关、面版数据相关、随机系数相关、数据合并等,与之类似的潜在类别模型也有着广泛应用;多项Probit模型(multinomial probit model,MNP)具有极高的灵活性,但其复杂的模型设定与庞大的运算量大大制约了其应用范围。本文在研究新动向上介绍了4个重要的研究关注点:由多种经典模型形式相结合而成的复杂模型;面向RP/SP数据、定序、排序、多选等不同数据类型的适宜模型;基于各种受限理性选择策略的更为真实的模型;以及考虑选择的时空背景的模型。
This article takes the general principles and application values of the discrete choice model system as a departure point and summarizes the classical model forms with respect to their basic theories and typical applications. Important latest developments are also introduced. Multinomial logit (MNL) model is the basis of the discrete choice model system, with the advantages of simplicity, reliability, and easy implementation. However, it also has some inherent theoretic defects, which led to the need for more refined models. Nested logit model is usually used to deal with problems of correlation among alternatives, no-choice alternative, and data enrichment. Its more general form is the generalized extreme value (GEV) model system; mixed logit model is suitable for handling random preference and some kinds of correlation problems, such as correlation among alternatives, panel data, random coefficients, and data for enrichment. A similar model form named latent class model is also widely used. Multinomial probit (MNP) model is highly flexible. However, its application is limited due to the complexity of model specification and very high computation demands. With regard to the new development of discrete choice model system, four important areas are introduced. These include complex new models derived from the combination of classical models; models suitable for dealing with revealed preference/stated preference (RP/SP), ordered, ranked, and multiple choice data; models based on bounded rationality choice which is more close to reality; and models considering the spatiotemporal background of choice.