ejikeugba/serp: serp 0.2.4

The R package serp fits cumulative link models (CLMs) with the 'smooth-effect-on-response penalty (SERP)'. The cumulative model developed by McCullagh (1980) is probably the most frequently used ordinal model in empirical studies. However, the stochastic ordering property of the general form of the model poses a very serious challenge in most empirical applications of the model. For instance, unstable likelihoods with ill-conditioned parameter space are frequently encountered during the iterative process. serp implements a unique regularization method for CLMs that provides the means of smoothing the adjacent categories in the model. At extreme shrinkage, SERP causes all subject-specific effects associated with each variable in the model to shrink towards unique global effects. Fitting is done using a modified Newton’s method. Several standard model performance and descriptive methods are also available. See Ugba (2021) and Ugba et al. (2021) for further details on the implemented penalty.