Abstract

The detection of business performance is to find out the soundness of business performance of an enterprise before the enterprise runs into any crisis or goes bankrupt in order to guard against any disaster before it happens. Generally speaking, when carrying out predicative analysis on business performance, most researchers adopt financial warning or credit rating mode. The data used are generally from events that have already happened. This paper, however, adopts a constructed business performance detection model to facilitate discrimination of business performance before the occurrence of any disaster. In this paper, the financial statements and various financial ratios of TSEC/GTSM listed fourth-party logistics providers were collected as sample data and four differential prediction models were constructed for business performance prediction of fourth-party logistics providers. Our empirical results showed that, the combination of Z-score and FOAGRNN hybrid model has differential prediction capacity significantly superior to other models, and the generalized regression neural network (GRNN) model after being adjusted with fruit fly optimization algorithm can effectively improve its prediction capacity. 

Key words: Z-score, generalized regression neural network (GRNN), fruit fly optimization algorithm, particle swarm optimization, grey relational analysis.