Regression with L 1 regularization (lasso) method was compared to principal component regression (PCR) in Statistical Downscaling (SDS) modeling to predict monthly rainfall. SDS modeling uses ill-conditioned (high correlation/multicolliniear) covariates, which can be solved with selection or shrinkage methods. In this study, we used two GCMs with different characteristics as covariates (CMIP5 and GPCP version 2.2). The results shows that the lasso method gave better results (smaller RMSE and RMSEP) than PCR for GPCP version 2.2, and as good as PCR for CMIP5 covariates.

This paper present development of clusterwise regression with a data set that has gamma distribution. Clusterwise regression is a method that finds simultaneously an optimal member of data in k cluster and each cluster have the best regression model. Analysis of a simulated data set has also been presented for illustrative purposes. Gamma and normal distributions were used for distribution of responses scenario with different parameters. This simulation study is carried out by initializing the number of clusters, classify observations randomly as an initial partition, move observation to the cluster giving the smallest residual and re-estimate the regression model from final partition. This simulation showed that clusterwise regression is able to form partition according to the distribution of data, also to form the best generalized linear model with Gamma distribution and linear regression model.

No abstract