Application of Selected Machine Learning Methods to Companies' Insolvency Prediction

被引：0

作者：

Wyrobek, Joanna ^{[1
]}

机构：

[1] Cracow Univ Econ, Dept Corp Finance, Rakowicka St 27, PL-31510 Krakow, Poland

来源：

EUROPEAN FINANCIAL SYSTEMS 2018: PROCEEDINGS OF THE 15TH INTERNATIONAL SCIENTIFIC CONFERENCE | 2018年

关键词：

machine learning; bankruptcy prediction; corporate finance; SUPPORT VECTOR MACHINE; FINANCIAL DISTRESS; BANKRUPTCY PREDICTION; DECISION TREE; CLASSIFICATION; CLASSIFIERS; ADABOOST; ENSEMBLE; FAILURE; MODEL;

D O I：

暂无

中图分类号：

K9 [地理];

学科分类号：

0705 ;

摘要：

The purpose of the paper was to test the efficiency of various modern machine learning methods on the representative sample of Polish companies for the time period 2008 - 2017. The novelty factor in the paper is that it uses a representative sample of companies, which seems to improve the efficiency of the models and that the training sample and the validation sample include data from different time periods and different companies (the training sample data covered the period of 2008 - 2013 and the validation sample covered the period 2014 - 217). The hypothesis verified in the paper is [H1] that: the most efficient algorithms in bankruptcy prediction are: Gradient Boosting Decision Trees and Random Decision Forest.

引用

页码：839 / 848

页数：10

共 52 条

[1]

Aktan S., 2011, Invest. Manag. Financ. Innov, V8, P52

[2] Bankruptcy forecasting:: An empirical comparison of AdaBoost and neural networks [J].