Predicting Economic and Financial Performance through Machine Learning

Abstract

. The aim of this paper is to demonstrate the usefulness of supervised machine learning algorithms in predicting the profitability of Romanian companies applying International Financial Reporting Standards (IFRS), both by regression and classification methods. The algorithms used in this research are linear regression (LinR), logistic regression (LogR), decision tree (DT), random forest (RF), K-nearest neighbor (KNN), and multi-layer perceptron (MLP). The results showed that both methods can produce models with high accuracy in profitability prediction. Thus, for regression, the best estimates were generated by the MLP model, and for classification, by the RF model. These results can be used to obtain sustainable models for predicting economic and financial performance, with a major impact on the management decisions of companies.