Fundamental Fitting of the SST Data using Linear Regression Models

The complexity of the relationship between the response and covariates becomes challenging in a dynamic model. The sea surface temperature (SST) data can be modelled as a linear combination of several climate parameters. By linear regression model (LRM) fitting we proposed two class models to investigate the effects of time covariates in the SST data modelling. Two time covariates seasonal and annual along with three continuous covariates are included in the second class model and the first class model consists of three continuous covariates. We observed that the time covariates have large influence in the model fitting of SST data. Meanwhile, transformation of rainfall in both class models improved the fit models. Our experimental comparisons of both class models reveal that an increase in R-squared, F-value and a reduction in residuals can be achieved by inclusion of the time covariates in the model. The preliminary statistical evidences of the linear models indicate that the proposed second class model fitting for SST data can used as fundamental fitting.