The Ultraviolet Index Classification Using Logistic Regression and Random Forest Methods for Predicting Extreme Conditions

EasyChair Preprint no. 10669

Abstract

The Ultraviolet (UV) Index is a crucial parameter for assessing the risk of harmful solar radiation exposure. Accurate prediction of extreme UV conditions is essential for public health management and environmental monitoring. This article presents a study focused on developing and evaluating a classification model using logistic regression (LR) and random forest (RF) methods to predict extreme UV conditions based on the UV Index. The objective of this research is to assess the performance of logistic regression and random forest algorithms in accurately classifying extreme UV conditions. Historical UV Index are used to train and validate the classification models. Data obtained from measurement of UV A and UV B radiation using a radiometer in Palu City, Central Sulawesi throughout 2022. This test uses split data randomly with a ratio of 70% for training data and 30% for testing data. Performance metrics such as accuracy are employed to evaluate the models. The results indicate that both logistic regression and random forest algorithms show promising performance in classifying extreme UV conditions. The logistic regression model achieves an accuracy of 0.958 while the random forest model achieves an accuracy of 0.997. The random forest algorithm has worked better than logistic regression. These findings demonstrate the potential of these models for accurately predicting extreme UV conditions. Furthermore, the model contributes to environmental monitoring efforts by providing insights into distribution of extreme UV conditions.

Keyphrases: logistic regression, Random Forest, ultraviolet

@Booklet{EasyChair:10669,
  author = {Alfin Syarifuddin Syahab and Galih Langit Pamungkas and Saif Akmal},
  title = {The Ultraviolet Index Classification Using Logistic Regression and Random Forest Methods for Predicting Extreme Conditions},
  howpublished = {EasyChair Preprint no. 10669},

  year = {EasyChair, 2023}}