Sector-Specific Financial Failure Prediction Model: Agriculture, Forestry and Fisheries Sector

In this study, it is aimed to create a sector-specific prediction model in order to eliminate the problems arising from creating a common model for businesses operating in different sectors in the financial failure literature. In this direction, prediction models were created by using the data of businesses operating in the Agriculture, Forestry and Fisheries sector between 2009 and 2019, taking into account the dynamics specific to the sector. In order to determine these dynamics, Random Forest, Stepwise Forward Backward Feature and K-Nearest Neighbors variable reduction methods were used and the performances of five different definitions of financial failure were evaluated. As a result of the evaluation of these definitions by Random Forest, Logistic Regression, Artificial Neural Network and KNearest Neighbors classification methods, the definition of FF_5, which includes the concepts of profitability and net working capital, differs from other definitions. As a result of the RF method applied to the model, 96.5% accuracy, 94.5% accuracy and 99% accuracy rates were calculated.