One of the significant problems of banks and investors in Iran is the lack of precise awareness about the financial performance of each bank and the roadmap for improving the conditions. Besides, the undesirable status of the financial performance of banks becomes evident only when the improvement of conditions is complicated. In this paper, a data mining-based early warning system (EWS) model has been presented to capture the financial performance of banks. To design this model, the CHAID decision tree has been used. Using this model, the banks have been classified as poor, medium, and good regarding financial performance, and the roadmap to achieving the desirable status has been determined. For this purpose, 13 Iranian banks have been investigated within the years 2003-2017. Eventually, the results obtained from the decision tree have been compared with the findings achieved from the CAMELS model. Based on the designed decision tree, 8 profiles have been extracted; 2 representing good, 3 medium, and 3 poor financial performance. Based on these profiles, according to the latest reports published by the studied banks, eight banks have a mediocre financial performance while five banks suffer poor financial performance. According to these profiles, four variables of the asset to shareholders’ equity ratio, the shareholders’ equity to loans ratio, the long-term debt to equity ratio, and liquidity coverage ratio were identified as the most relevant variables associated with the financial performance of banks.
