Combining Two Classification Methods for Predicting Jakarta Bay Seabed Type Using Multibeam Echosounder Data

Abstract

Classification of seabed types from multibeam echosounder data using machine learning techniques has been widely used in recent decades, such as Random Forest (RF), Artificial Neural Network (ANN), Support Vector Machine (SVM), and Nearest Neighbor (NN). This study combines the two most frequently used machine learning techniques to classify and map the seabed sediment types from multibeam echosounder data. The classification model developed in this study is a combination of two machine learning classification techniques, namely Support Vector Machine (SVM) and K-Nearest Neighbor (K-NN). This classification technique is called SV-KNN. Simply, SV-KNN adopts these two techniques to carry out the classification process. The SV-KNN technique begins with determining test data by specifying support vectors and hyperplanes, as was done on the SVM method, and executes the classification process using the K-NN. Clay, fine silt, medium silt, coarse silt, and fine sand are the five main classes produced by SVKNN. The SV-KNN method has an overall accuracy value of 87.38% and a Kappa coefficient of 0.3093.