Spatial co-location pattern mining of facility points-of-interest improved by network neighborhood and distance decay effects

Abstract

The aim of mining spatial co-location patterns is to find the corresponding subsets of spatial features that have strong spatial correlation in the real world. This is an important technology for the extraction and comprehension of implicit knowledge in large spatial databases. However, existing methods of co-location mining consider events as taking place in a homogeneous and isotropic context in Euclidean space, whereas the physical movement in an urban space is usually constrained by a road network. Furthermore, previous works do not take the 'distance decay effect' of spatial interactions into account, which may reduce the effectiveness of the result. Here we propose an improved spatial co-location pattern mining method, including the network-constrained neighborhood and addition of a distance-decay function, to find the spatial dependence between network phenomena (e.g. urban facilities). The underlying idea is to utilize a model function in the interest measure calculation to weight the contribution of a co-location to the overall interest measure instance inversely proportional to the separation distance. Our approach was evaluated through extensive experiments using facility points-of-interest data sets. The results show that the network-constrained approach is a more effective method than the traditional one in networkstructured space. The proposed approach can also be applied to other human activities (e.g. traffic accidents) constrained by a street network. ARTICLE HISTORY

Nonparametric Significance Test for Discovery of Network‐Constrained Spatial Co‐Location Patterns

“…For example, a road network skeleton partitioning scheme was proposed to construct transactions, and the statistical diagnostics was employed to detect network‐constrained spatial co‐location patterns (Tian, Xiong, and Yan ). The skeleton partitioning scheme may also cause the loss of neighbor relationships among objects across adjacent roads (Yu et al, ). To alleviate this problem, the neighbor relation R over instances in I can be defined as a network‐constrained neighborhood by using the shortest path distance on the network space, and then the participation index can be extended to measure the prevalence of network‐constrained spatial co‐location patterns (Yu ).…”

“…Although some statistical methods can also be used to discover co-location patterns without user-specified prevalence thresholds, for example, cross K-function (Ripley 1976), cross nearest-neighbor distance method (Okabe and Miki 1984), and statistical diagnostics (Sierra and Stephens 2012), these methods only measure pairwise co-location of spatial features and cannot be easily extended to measure spatial interactions among more than two measures (Huang et al 2004;Yu et al 2016). Barua and Sander (2014) developed a significance test for evaluating the prevalence of co-location patterns.…”

Redes de regras de associação filtradas e multialvo

“…Para extração das regras, os Grafos foram utilizados como estruturas de apoio para que os itemsets fossem formados e as medidas de suporte e confiança calculados. Destacam-se nessa fase as pesquisas de Wang e Xu (2018), por estabelecer Regras de Associação diretamente ligadas à estrutura do Grafo, o trabalho de Yu et al (2017), que apresenta a extração de padrões espaciais em Redes e a pesquisa de Namaki et al (2017), por estabelecer regras de associações temporais com o uso de Redes de eventos heterogêneas.…”

“…Quanto às tarefas de agrupamento (VALLE; RUZ; MORRÁS, 2018;ZHANG et al, 2018;GUO;YU et al, 2017;TSAI et al, 2017;FU et al, 2017;MARTÍNEZ-BALLESTEROS et al, 2017;SLAIMI et al, 2017;MONDAL;BHATTACHARYA;MONDAL, 2016;ZHENWEI;LINGYUN;LIZHU, 2016;RUSH et al, 2016;XING, 2016;LI-XIONG et al, 2015; RODRIGUES; GAMA; FERREIRA, 2012; AL-KHASSAWNEH; BAKAR; ZAINUDIN, 2012; RAJPUT; THAKUR; THAKUR, 2012; CHOOBDAR; SILVA; RIBEIRO, 2011; ALZOUBI; OMAR; BAKAR, 2011), várias abordagens fazem uso das Redes na etapa de pós-processamento da mineração. Destaca-se os trabalhos de Karimi-Majd e Mahootchi (2014) que efetua, na etapa de pós-processamento da mineração, o uso de Redes para auxílio no agrupamento e mapeamento das regras.…”

“…As Redes também podem ser utilizadas a fim de otimizar a formação dos itemsets frequentes, reduzindo a quantidade de itens a serem explorados pelo usuário. Um dos desafios das técnicas de Mineração de Regras de Associação é que elas geralmente geram um grande número de "padrões", o que torna extremamente difícil para um pesquisador decidir quais regras são confiáveis e valem a pena serem analisadas (ZAKI;MEIRA, 2013). Apesar de diversos trabalhos realizados buscarem alternativas para extração do conhecimento de forma objetiva, a utilização das Redes no processo de ARM, com o uso de um atributo alvo, têm sido pouco explorada na literatura.…”