Özet:

The determination of the company’s risk profile is studied by the concepts of company failure in literature. This topic, especially after the 1929 great breakthrough, was very important. At first, while focusing on the differences between the financial indicators of risky and successful companies, in particular parallel to the developments in computer technologies, in recent years, computer systems have become one of the most important components in the company failure forecast. In particular, with the beginning of machine learning methods to be used in this field, the company’s failure has been significantly predicted. In this study, the construction material wholesale that has been operating for 38 years in the Erzurum province has been addressed in the framework of the company’s failure and the method of learning a machine appropriate to the company’s failure. The problem-affected variables are identified by the Basic Components Analysis (TBA). In recent years, the applicability of hybrid models created with the use of artificial nerve networks (YSA) and support vector machines (DVM) along with the TBA method in machine learning has been studied and the predictive performance has been compared to simple YSA and DVMs. The predictive success of the hybrid models used in combination with TBA has been shown to give more satisfactory results than the simple YSA and DVMs. In particular, the conclusion has been made that the TBA-DVM model can be effectively used as an alternative method in the prediction of company failure.

Anahtar Kelimeler: