Clustering Analysis of a Spatiotemporal Dataset with a Novel Kernel Density Estimator
Journal
Proceedings - International Conference on Machine Learning and Cybernetics
ISBN
9798350303780
Date Issued
2023-01-01
Author(s)
Abstract
A vast number of spatiotemporal datasets collected from a wide range of sources has motivated scientists to develop effective approaches to identify interesting patterns hidden in these datasets. In this respect, kernel density estimators, which belong to a class of non-parametric estimators in statistics, have been widely exploited in recent years. With this background, we have developed a novel kernel density estimator aiming to provide accurate analysis results. According to the evaluation with a real spatiotemporal dataset, which collected emergency medical service records in a county in the United States, the proposed kernel density estimator can approximate the probability density function significantly more accurately than a conventional kernel density estimator. Furthermore, we have exploited the proposed kernel density estimator to identify interesting patterns hidden in the real spatiotemporal dataset.
Subjects
Clustering analysis | Kernel density estimation | Spatiotemporal data
Type
conference paper