Variable selection in gamma regression models via artificial bee colony algorithm

Improving the Accuracy of Rainfall Prediction Using Bias-Corrected NMME Outputs: A Case Study of Surabaya City, Indonesia

Generating an accurate rainfall prediction is a challenging work due to the complexity of the climate system. Numerous efforts have been conducted to generate reliable prediction such as through ensemble forecasts, the North Multi-Model Ensemble (NMME). The performance of NMME globally has been investigated in many studies. However, its performance in a specific location has not been much validated. This paper investigates the performance of NMME to forecast rainfall in Surabaya, Indonesia. Our study showed that the rainfall prediction from NMME tends to be underdispersive, which thus requires a bias correction. We proposed a new bias correction method based on gamma regression to model the asymmetric pattern of rainfall distribution and further compared the results with the average ratio method and linear regression. This study showed that the NMME performance can be improved significantly after bias correction using the gamma regression method. This can be seen from the smaller RMSE and MAE values, as well as higher R² values compared with the results from linear regression and average ratio methods. Gamma regression improved the R² value by about 30% higher than raw data, and it is about 20% higher than the linear regression approach. This research showed that NMME can be used to improve the accuracy of rainfall forecast in Surabaya.

A New Ridge-Type Estimator for the Gamma Regression Model

The known linear regression model (LRM) is used mostly for modelling the QSAR relationship between the response variable (biological activity) and one or more physiochemical or structural properties which serve as the explanatory variables mainly when the distribution of the response variable is normal. The gamma regression model is employed often for a skewed dependent variable. The parameters in both models are estimated using the maximum likelihood estimator (MLE). However, the MLE becomes unstable in the presence of multicollinearity for both models. In this study, we propose a new estimator and suggest some biasing parameters to estimate the regression parameter for the gamma regression model when there is multicollinearity. A simulation study and a real-life application were performed for evaluating the estimators' performance via the mean squared error criterion. The results from simulation and the real-life application revealed that the proposed gamma estimator produced lower MSE values than other considered estimators.