Advances and Applications in Statistics
Volume 60, Issue 1, Pages 79 - 91
(January 2020) http://dx.doi.org/10.17654/AS060010079 |
|
ON COMPARING THE EFFICIENCY OF TWO CLASSIFICATION METHODS FOR RECIPITATION OCCURRENCE IN BANGKOK
Wandee Wanishsakpong, Khairil Anwar Notodiputro and Jeeraporn Thaithanan
|
Abstract: This study is aimed at describing factors affecting precipitation occurrence in Bangkok during 2006-2016. Daily rainfall data were acquired from the Thai Meteorological Department and used in this study. Two classification methods, namely, the discriminant analysis and logistic regression were used to classify factors affecting precipitation occurrence in Bangkok. The Apparent Error Rate (APER) was employed to assess the efficiency of the methods. A method which provides results with smaller error has been considered as a more effective method. The dependent variable used for this study was the occurrence of precipitation which was a binary response variable. The explanatory variables consisted of four variables, namely, relative humidity, temperature, wind speed and wind direction. The results of this study indicated that using the discriminant analysis, it was found that factors affecting the precipitation were relative humidity, wind direction and wind speed with an APER of 24.46%. On the other hand, the logistic regression analysis showed that factors affecting the precipitation were relative humidity, temperature, wind direction and wind speed with an APER of 23.64%. It is clear that the logistic regression analysis has been slightly more effective to classify the occurrence of precipitation when compared to the discriminant analysis. |
Keywords and phrases: APER, classification, discriminant analysis, logistic regression.
|
|
Number of Downloads: 294 | Number of Views: 1041 |
|