Please use this identifier to cite or link to this item: http://hdl.handle.net/10397/88297
Title: Spatial big data analytics of spatiotemporal mobility characteristics of the elderly
Authors: Shi, Zhicheng
Degree: Ph.D.
Issue Date: 2020
Abstract: Many countries worldwide have ageing populations. In 2019, the size of the global population of the elderly who aged 65 years or over was 703 million. As the number of the elderly population increases, they will need more care and help. People's basic needs include clothing, food, housing and transport, and this is just as true for the elderly. The thesis focused on study the elderly behavior, which is still an open issue. Specifically, the thesis considered the spatiotemporal and mobility characteristics of the elderly and took Beijing as a study area. Existing research based on annual traffic surveys or questionnaires on the spatial distribution of the elderly lacks near real-time prediction (e.g., next hour), large sample sizes (in the millions), precise location (latitude and longitude coordinates), and high frequency (hourly) data. Research based on real-time spatial big data, such as smart card data, can fill this research gap. The aim of this thesis was to understand the spatiotemporal and mobility characteristics of the elderly by using smart card data. This aim was achieved by realizing the following interrelated research objectives: 1. To develop a Voronoi construction method based on an integrated clustering method for region partition. 2. To study the spatial distribution characteristics and mobility behavior of the elderly. The spatial distribution characteristics include a) identifying their home locations using smart card data, b) explaining why the elderly distribution like this, and c) detecting places they visit frequently. The mobility behavior include a) travel time, b) travel distance, c) travel duration, and d) travel frequency. A data-driven methodology was adopted for this thesis. Smart card data for Beijing, a city with a high life expectancy, were used in the analysis.
To achieve objective 1, an integrated method was developed to detect clusters in datasets with multiple densities and shapes features. Two improvements were made to the classic clustering methods: a) cluster number was estimated automatically, and b) only one parameter was required. With these improvements, multiple densities and shapes of clusters could be detected effectively. The clustering method was also scalable for different kinds of dataset. To achieve objective 2, three targets were set to examine and determine the spatial distribution patterns of the elderly by means of smart card data. First, the spatial distribution of the elderly population in a city was analyzed using the Voronoi diagram which is based on an integrated clustering method. The proposed method can efficiently detect clusters with multiple densities and shape and accurate to present the spatial distribution of elderly. Second, the spatial distribution pattern of the elderly was measured and explained by a newly proposed model of PoI-based elderly livability index computed based on weighted factors including restaurants, parks, hospitals, shops and bus stops. Third, the spatial connectivity between regions in which the elderly travels was used to describe where the elderly frequently travels. A quasi-gravity model was developed to reveal the relationship between spatial connectivity and the PoI-based elderly livability index. Three important findings were yielded: a) the spatial distribution of the elderly's home locations shows clear clustering characteristics; b) the spatial distribution of the elderly has a strong relationship with public service facilities, such as restaurants and hospitals; and c) the connectivity of pairs of regions is related to the distribution of public facilities in the connected regions. To understand the mobility behavior of the elderly, two methods were adopted: (a) quantitative analysis of spatiotemporal travel behavior by estimating the parameters of travel patterns and the subsequent presentation of such behavior graphically, and (b) discovery of the distribution functions of the travel characteristics, both by function curve fitting and by testing the goodness of fit of the identified distribution functions. A number of important findings have yielded on elderly mobility behaviors in megacities: (a) most of the elderly's travels are approximately 1 km and the median distance is 4 km, which is shorter than the adults, and their travel distance follows an exponential function, unlike the travel distance distribution of adults, which follows a Gaussian function; (b) most of the elderly travel for 4 minutes, which is half the time of the adults' travels, and travel time follows a Gaussian distribution; (c) the elderly's travel departure time has a morning peak at 9:00 am (compared with 8:00 am for adults) and no clear peak in the afternoon; and (d) most of the elderly travel once per day, which is the same as the adults. The significance of this thesis lies in the series of new analytics methods developed and the comprehensive findings regarding the spatial distribution patterns and mobility behavior of the elderly in megacities. These findings add new knowledge to the field. The new methods could be widely applied in urban planning, management and services for the aging population.
Subjects: Older people -- Orientation and mobility
Older people -- Transportation
Spatial analysis (Statistics)
Hong Kong Polytechnic University -- Dissertations
Pages: x, 141 pages : color illustrations
Appears in Collections:Thesis

Show full item record

Page views

8
Citations as of May 22, 2022

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.