Location:Home > Engineering science > Computer Science > Computer technology > Company Financial Early-warning System Based on Machine Learning
Details
Name

Company Financial Early-warning System Based on Machine Learning

Downloads: []
Author
Tutor: ZuoFei
School: Xiamen University
Course: Computer technology
Keywords: Financialearly-warning,unbalanced datasets classification,combinationclassifier
CLC: TP181
Type: Master's thesis
Year:  2014
Facebook Google+ Email Gmail Evernote LinkedIn Twitter Addthis

not access Image Error Other errors

Abstract:
Risk is everywhere. the cases that our private enterprises go to bankruptcy because ofmoney problem often happen.The financial crisis of the company threatens the survival and development of the enterprise.Therefore, the financialcrisis early-warningsystem for the enterprisefinancial riskis very necessary.In terms of choosing good financial crisis early-warning features, the features that reflect the areas of the company--the ability to grow、 cash flow、 the ability to operate、 debt paying ability、 capital formation、 the profitabilityconstitute the preliminary selected early-warning features in the paper.then,the features are filtrated scientifically by using the methods of the information gain significant test and correlation test principal component extraction.In terms of building the early-warning model,For the imbalance of financial normal and abnormal financial class sample set, this paper introduces semi-supervised learning mechanismand builds an integrated learning model based on improved KNN-SVM’s model.the experimental shows that the model constructed in this paper has a high predictive ability.The innovations of this paper are:Firstly, the preliminary selected early-warning features are all-sided and the final early-warning features are reasonable and scientific. Secondly, the semi-supervised learning mechanism is introducted to increase minority class classification knowledge, so the predictive ability of the model ishigher. Thirdly, estimation of distribution algorithm is applied to optimize the model parameters. Fourthly, fuzzy sets are introductedinto the system.we can know the degree of confidence of the sample predicting outcomes.
Related Dissertations
Last updated
Sponsored Links
Home |About Us| Contact Us| Feedback| Privacy | copyright | Back to top