The Performance Forecasting Study On The Listed Enterprises Assets Restructuring In Tourism Firms

As we know, Assets restructuring plays an important role in improving the asset structure and distributing the resources. It has been frequently selected by many firms. It has also been a tool can well change the performance. Even though a lot of companies had a better development, but there still so many firms failed by choosing restructure. Does it really raise the performance? Researchers can’t agree with a consistent conclusion. Now the wave of assets restructuring has been constantly raised, As a way to judge the success of assets restructuring, performance forecasting seems to be very important. So the performance forecasting study on the listed enterprises assets restructuring has become more and more urgent..This paper based the dates on tourism enterprises in China during the year 2000 to year 2013.Used the Kolmogorov-Smirnov test、T-test or Kruskal-Wallis、relative analysis and collinearity test to choose the finical index we need.To judging the performance of assets restructuring,we use four single traditional prediction model methods(MDA、Logit、Probit、DT)and two artificial intelligence methods(CBRN SVM) to forecast the performance.But we found that the successful samples has not equal to the failures, thence we put out an improved smote method (NT_smote). Then we use the six methods to predict the performance from the untreated and improved samples. Comparing the experiment results, we found that it has significantly raised the prediction accuracy after improving.Secondly, although we had successfully improved the accuracy from the unbalanced dates, but the results to failure samples were too low. So,under the foundation of increasing,we propose the method of clustering weighted combination, then we improved the six single methods,thereby it produced five new methods (CWCLogit、CWCProbit、CWCDT、CWCCBR. CWCSVM). Finally we had proved that the first four methods:CWCLogit、CWCProbit、CWCDT、CWCCBR are valid.At last, we summarized the results come from the method of NT_smote and clustering weighted combination. Given out some relevant management Implications and put out the future research directions.