Original scientific paper

https://doi.org/10.17559/TV-20160913205831

Data Mining Approach in Climate Classification and Climate Network Construction – Case Study Montenegro

Savo Tomović orcid.org/0000-0002-9388-7784 ; University of Montenegro, Džordža Vašingtona bb, 81000 Podgorica, Montenegro
Predrag Stanišić ; University of Montenegro, Džordža Vašingtona bb, 81000 Podgorica, Montenegro
Srdjan Kadić ; University of Montenegro, Džordža Vašingtona bb, 81000 Podgorica, Montenegro

Abstract

In this study, we present results of applying data mining techniques on meteorological dataset obtained from the Institute of Hydrometeorology and Seismology of Montenegro. The dataset covers the measurements taken from all 11 main meteorological stations in Montenegro for the period 2010-2015. We build new climate classification system based on decision tree. The system is simpler (i.e. uses fewer attributes) and more accurate than the well-known Köppen climate classification system. In addition, we propose a novel procedure for climate network construction. Finally, we identify the regions within the same climate type in Montenegro’s climate network with the Girvan-Newman algorithm for community detection and achieve better results with respect to classical K-means and hierarchical clustering algorithms.

Keywords

clustering; climate networks; Koppen climate classification system; mining meteorological data

Hrčak ID:

204450

URI

https://hrcak.srce.hr/204450

Publication date:

20.8.2018.

Visits: 1.803 *