Ability Comparison Adaboost Learning Machine Algorithms and Probable Classification Bayesian in Predicting Managers Over Confidence Iranian Capital Market Companies

Subject Areas : Financial engineering

shokoufeh etebar ¹ , Roya Darabi ² , Mohsen Hamidian ³ , Seiyedeh Mahbobeh Jafari ⁴

1 - PhD student of accounting, , Faculty of Economics and Accounting, Islamic Azad University, South Tehran Branch, Tehran, Iran
2 - Faculty Member, Faculty of Economics and Accounting, Islamic Azad University, Tehran South Branch, Tehran, Iran
3 - Faculty Member, Faculty of Economics and Accounting, Islamic Azad University, Tehran South Branch, Tehran, Iran
4 - Faculty Member, Faculty of Economics and Accounting, Islamic Azad University, Tehran South Branch, Tehran, Iran

Received: 2018-05-24 Accepted : 2018-09-05 Published : 2019-06-22

Keywords: Management over Confidence, Adaboost Learning Machine Algorithm, Probable Classification Bayesian Algorithm,

Abstract :

More confidence is a personality trait, it may affect the decisions of corporate executives, and Management with irrational behavior can affect the company's performance in the long run. The main purpose of this applied research is “Ability comparison Adaboost Learning Machine Algorithms and Probable Classification Bayesian in predicting Managers over Confidence Iranian Capital Market Companies during the years 2009 to 2017 ". In this regard, Adaboost artificial intelligence algorithms and Probable Classification Bayesian were proposed to validate this year and the following year in order to provide a model for predicting Management over Confidence in Tehran Stock Exchange companies. In total, 34 different industries were selected as statistical samples (financial data of 3145 companies-year in total). The method of data collection in the theoretical part of the library method and for calculating and statistical classification of financial data, from software's Excel and to test the research hypotheses Used from the software matlab 2017. The research findings show that, obtained nonlinear models prediction of the Adaboost algorithm and the probable Bayesian algorithm, both have ability to predict the over confidence of management for this year and one year later, But Adaboost 's prediction model, in contrast to the Bayesian's prediction model, has better results in predicting Management over Confidence. Which indicates greater power in learning and better performance of this model in order to predict the uncertainty of management.