clustering algorithm

The use of traditional forecasting tools and methods has a high error and has a poorer performance compared to newer methods and nonlinear models. One of the most widely used methods and algorithms in predicting the use of machine learning. The main purpose of this study is to investigate the application of machine learning in providing a model for predicting the bankruptcy of 308 companies listed on the Tehran Stock Exchange in the period 1389 to 1398 (3080 years - company) to test the hypotheses of multiple regression of composite data. In order to implement the Medians-K clustering algorithm and related calculations, R statistical calculation software was used. The results show that among the financial ratios identified in the first model, only the ratio of net income to total assets and the ratio of market value of equity to total market value can improve the ability of the Altman bankruptcy prediction model. Also, in the second model, the specified financial ratios have the ability to improve the bankruptcy forecast model, and by adding the Devscore variable for groups based on industry and size, the modified model improves the bankruptcy forecast, The results shows that a company is more likely to go bankrupt if it has bankruptcy-related financial ratios that are lower than the average of its cluster peers..

Abstract Machine learning is a broad discipline that has designed learning algorithms that can guide stimuli, detect spoken language, and discover hidden adjustments in data volume growth, of which financial data is no exception. Therefore, the purpose of this study is to investigate the application of machine learning in providing a model for detecting accounting distortions in companies listed on the Tehran Stock Exchange. Statistical analysis of the research was performed based on the extracted data of 308 companies listed on the Tehran Stock Exchange in the period 1389 to 1398 (3080 years - company) and the screening method was used for sampling. Accounting distortions of the dependent variable obtained through the virtual variable zero and one and the variables of non-discretionary accruals, change in accounts receivable, change in inventory, soft assets, change in cash sales, change in return on assets and issuance of securities as Independent variables are considered and nonlinear regression is used to test the hypotheses and R statistical calculation software is used to implement the Medians-K clustering algorithm and related calculations.