Voronoi classified and clustered data constellation: A new 3D data structure for geomarketing strategies

Abstract

Location plays a very important role in geomarketing. Location tells where the customers are, identifies something in the surrounding area or solves problems regarding the location of a new outlet. However, in an urban area, the locations have a vertical component due to high-rise and multilevel buildings. This situation requires a new approach that can handle three-dimensional data for location analysis. In this research, a novel 3D data structure is introduced to manage and constellate locations in three-dimensional space. The data structure is designed based on a group of classifications and clusters, and supplemented with the additional element of nearest-neighbour information. The locations are analysed to determine a geomarketing strategy by using several methods, such as single-nearest-neighbour, k-nearest-neighbour (kNN) and reverse-k-nearest-neighbour (RkNN) analyses. These analyses are performed based on encoded neighbour information of the Voronoi diagram that is extracted from the data structure. From the results, various tasks pertaining to geomarketing strategy can be carried out, such as identifying nearby competitors, locating target customers for marketing purposes and analysing the impact of opening a new outlet on competitors. Additionally, the proposed method is tested for its ability to handle large amounts of geomarketing data in terms of its efficiency in time retrieval and storage. The data structure is compared with 3D R-Tree to analyse its performance and efficiency. 3D R-Tree is chosen because it is the most commonly used structure in spatial databases. The test demonstrates that the proposed method requires the least amount of Input/Output than 3D R-Tree. The performance of the data structure is also evaluated; the results indicate that it is outperforms it competitors by responding 60–80% faster to query operations.