Research on Liquidity Risk Identification Based on KNN

Xie Jiafang

Abstract:Enterprise being caught in liquidity risk and capital chain rupture and then going bankrupt is a common problem in the enterprise management.Therefore, studying the liquidity risk of the enterprise has important theoretical and practical significance.In this paper, we analyzed the early warning and the cause of the Enterprise liquidity risk.We selected 38 ST listed companies of manufacturing industry in Shanghai and shenzhen stock exchange between 2012—2015, and also randomly selected 38 normal status of manufacturing listed companies to match the study.Firstly, we adopted Chinese scholar Zhou Shouhua's prediction model-F fraction model which are modified on z-score model after the establishment of a new financial crisis to evaluate the liquidity risk of the enterprise, and the results of the assessment can be divided into two categories. Additionally, we can judge the shortfall caused by enterprise are either operational or from long-term investment and financing according to enterprise's daily operational activities. Therefore we can extract eight dominating factors which will lead to enterprise liquidity risk.We quantify these eight indicators and use the entropy weight method to weight them. Finally,70% of the sample data is used as a training set, and 30% of the sample data used as the forecast set.Then we use and the k-neighbor analysis method based on the importance of variables and the k-neighbor method based on the observed similarity to predict the sample data.At the same time, we use the C4.5 and C5.0 algorithm in the decision tree model to predict the sample data so as to compare this two model.The results showed that the“train”function in the K neighbor method has the best prediction result.

Key Words:Liquidity Crisis; Z-score Model; Entropy Method; K Neighbor Method;The Decision Tree Model