Comparison between the Naïve Bayes and Hierarchical Clustering to Classify The Global Landslide Catalog for the Prediction of the Landslide.

Abstract

Machine Learning has been used since long to identify the features of a given datasets that are important for the prediction. Landslides are complex events taking place in the various regions of the world. It is the movement of the debris, soil or rocks from an upper plane in downward direction. Identification of the features that are used for the Landslide involves consideration of various categories of parameters. Present paper studies about the performance comparison between a supervised algorithm Naïve Bayes and unsupervised algorithm Hierarchical Clustering. Naïve Bayes is a non parametric supervised algorithm that can be used for the forecasting purposes in the field of Agriculture, Economics, Aviation etc, whereas Hierarchical Clustering is used to partition the available instances of a dataset into optimal homogeneous groups on the basis of the similarities between the datapoints. The present paper draws a comparison between the accuracy of the Naïve Bayes and Hierarchical Clustering for the prediction of the Landslide dataset. The dataset used is the Global Landslide Catalog that has important parameters like date, location coordinates, country, trigger of the event, continent etc. Before the implementation of both the algorithms, reduction of the parameters is carried out using subset evaluation of the parameters and considering only the most important.