Prediction of financial distress: A modelling approach for construction companies in Turkey

This thesis aimed to develop a logistic regression model for predicting financial distress in Turkish construction companies using their financial information. Based on financial information from the last 12 months, the model can predict with high accuracy whether the companies in the sample would be in financial difficulties. The model employed independent variables related to liquidity, leverage, efficiency, and profitability. It demonstrated satisfactory performance, with an overall accuracy of 86.54% and a sensitivity rate of 96.15%. The model's practical applications can be in credit risk assessment, risk-based pricing, and government tender processes. Future research recommendations include expanding the dataset, comparing different machine learning algorithms, and incorporating non-financial factors to improve predictive capabilities.