Compare failure prediction models based on feature selection technique: empirical case from Iran

Abstract

Due to the uncertainty of the current business environment and global competition, not only the failure prediction of corporations is very important, but also it is so crucial to prevent the wasting of individual’s capital. Failure of corporation does not take place at overnight. Warning signals of a company heading for business failure emerge much earlier than the actual failure. Therefore, these signals could be used as input variables in prediction models, to predict the failure in advance. Along with the advancement of artificial intelligence and database technology, data mining techniques are applied universally to management problems such as failure prediction of corporations since 1990s. The most used models for failure prediction are the neural networks. A huge amount of information about the corporations that derived from financial reports could be used to determine the failure of companies, but it needs much time and human resources. Selection of financial variables or features is very important stage in process of business failure prediction. In this study, 2 prediction models would be compared with each other in 3 stages. These models are neural networks that named “MultiLayer Perceptron”. One of these models is trained with original dataset and the other one is trained with a dataset contained the selection of features of original data set. The comparison of prediction models is based on accuracy of them and the type I and II errors.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 12 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.