Optimising Maritime Big Data by K-means Clustering with Mapreduce Model

Abstract

During the management and operation, the maritime industry has collected a large amount of data in marine navigation, which has posed a great challenge in terms of resource saving (memory and processing capacity) and utility efficiency. Therefore, the highly specialised nature of the marine navigation and the maritime big data must be analysed to assist the scientists and operational engineers to extract the useful information from this data using algorithms with big data platforms. However, a specific model for big data application, which has a lot of methods for performing such as data visualisation techniques, machine learning, deep learning, etc., has not been extensively studied in the field of marine navigation to provide adequate comparisons. In this paper, we apply Mapreduce (MR) model to the big data of marine navigation. Particularly, we use a standard clustering algorithm called K-means based on the MR model to process the data of marine traffic in the South Vietnam Sea region. According to the main results obtained, we consider making the inference or the prediction of the clustering data which is collected and shown the dashboard of maritime ships traffic, including the scale, the spatial and time-series distribution situation.