scroll to top
Stuck on your essay?
Get ideas from this essay and see how your work stacks up
Word Count: 4,425
AbstractThe potential of a company going bankrupt in the near future is a crucial aspect when making business decisions Although this topic is widely studied typically financial data is used to make predictions This leads to a problem that the predictions are not up-to-date as financial statements are published long after the business year This study investigates if using company board members formal network graph metrics evolution can affect bankruptcy prediction accuracy as this approach is not yet widely researched and changes in the formal network are available without notable delay More specifically this study compares different machine learning methods to find the best method among them The research reveals that network metrics especially PageRank degree and eccentricity indeed improve bankruptcy prediction models The best performing model was a decision tree based random forests that achieved best results using six months of data and classifying bankruptcy up to nine months in advanceIntroductionBankruptcy prediction of companies especially banks is a well-researched area since the late 1960s 1 It is an important problem since it can have a high influence on business decisions and profitability In fact the forecast of bankruptcies is necessary for different types of public and commercial organizations as a failed business can cause failures of other companies and affect the rest of the financial system and shifting bankruptcy rate can indicate changes in the economic environmentThe most common approach used before is using companies financial data to forecast bankruptcy The techniques to predict bankruptcy used in the past are divided into two extensive categories linear and non-linear methods For example the linear techniques that have been used include linear discriminant analysis 1 multivariate discriminate analysis 2 and logistic regression logit 9 There are some limiting presumptions when using linear statistical methods such as the linearity normality and independence amongst predictor or input variables Statistical methods can have problems with effectiveness and validity because violation of these presumptions for independent variables frequently occurs with
@Kibin is a lifesaver for my essay right now!!
- Sandra Slivka, student @ UC Berkeley
Wow, this is the best essay help I've ever received!
- Camvu Pham, student @ U of M
If I'd known about @Kibin in college, I would have gotten much more sleep
- Jen Soust, alumni @ UCLA