The Study On Information Forewarning Of Companies' Financial Risk Based On Data Mining

With the rapid development of economic and social Chinese,complexity and uncertainty characteristics show the economic aspects of the obvious,the manufacturing industry enterprise financial risk or even bankruptcy situation is increasing.As a branch of the manufacturing industry,the chemical industry faces many internal and external challenges,therefore,how to dig many financial the formation of enterprise data,and then found the relevant information for the financial risk early warning with utility,become a problem urgent to deal with.Just,sustainable development of perfect data mining technology to deal with this problem.The direction is the function of data mining for massive data processing,after the discovery of the laws with the knowledge hidden.Therefore,using this technology,the financial risk early warning of listed companies of the chemical industry to establish corresponding The efficient early-warning model has great value and function.This paper first introduces the basic theory and research results of financial risk warning and data mining,and introduces five kinds of data mining algorithms.The article,contact the domestic chemical industry listed companies own actual characteristics,establishes the corresponding index mechanism of financial risk warning.Finally in 2016 184 Sample Firms from 2011 to 2014 four from the data as the premise,the use of Weka data mining model,this technology will be applied to the financial risk early warning,and effective mining algorithm for various types of non-financial indicators,the influence on the financial risk early warning and timeliness of samples for the forecasting results are explored.From the results of research:first,in the five kinds of data mining algorithms,decision tree,the best prediction results of support vector machine and random forest of these three models,the choice for subsequent experiments.Second,after consideration of the influence of non financial index has the forecast function to introduce this kind of constructing index future the model is generally better than only using the financial indicators of the model.Third,the timeliness of the sample is very important for the early warning of financial risk,from the time of ST is closer to three,prediction model has a greater accuracy.Analysis of short period early warning perspective,three types of models have predictive effect good,but the long-term perspective,the prediction effect of support vector machine is superior to other models.