A FEW NOTES ON DEPLOYMENT OF SUPERVISED CORPORATE FINANCIAL DISTRESS PREDICTION MODELS IN SMALL ENTERPRISES

Prediction of corporate financial distress is often based on static classification models constructed using various supervised statistical methods, e.g. discriminant analysis, logistic regression, decision trees. Regardless of the selected classification method, a carefully checked data set containing information about the occurrence of financial distress and quantitative characteristics, most typically three or four years prior to the time of financial distress, of selected companies has to be available for a company in order to utilize this approach. Moreover, employees possessing at least moderate data analytic skills are vital to fit, interpret, validate, deploy and, eventually, upgrade models. These requirements make it quite infeasible for small businesses to construct such models on their own and utilization of freely available models from other sources is possibly preferred. In the paper we propose deployment of such models using Shiny, a web application framework for R. The deployed model is based on data obtained from CRIF - Slovak Credit Bureau, s.r.o. and we focus on simple automated interpretation of model outputs for a particular company.