Application Research On Business Valuation System In Retail Services

Business valuation is the basis and premise for mergers and acquisitions as well as other investment and management practice. However, existed business valuation models could not predict a company's profits in the future, thus greatly influence the effectiveness of valuation. So in this paper, we adopted an improved LIRBF correlated with discounted cash flow to design a business valuation system.Improved LIRBF mixed regression method overcomes the shortcomings exited in traditional regression method which usually automatically assume that the explanatory variable and the dependent variable are linear relationship in real applications. This method enhances non-linear of the model by increasing the radial basis. The model uses initial and radial basis function variables. Firstly, we determine the center, radius and the number of Gaussian radial basis function by genetic algorithms. Secondly, non-linear transformation of input variables obtained by the RBFs of the best individual in the final generation. Thirdly, we obtain coefficients by using the maximum likelihood estimation regression. Our paper works with the simulation of business valuation on China's listed retail companies. The improved LIRBF mixed regression models are found to be effectiveness when compared with the corresponding multi-logistic regression methods and the RBFGA method and LIRBF method correlated with discounted cash flow. Inventory turnover is commonly used to measure performance of inventory managers, the speed of cash flow and level of competitiveness of the retail business. Therefore, it is an important index for investors and managers. Not every company publishes it in practice. In the previous studies, we usually use the multi-linear regression method to calculate inventory turnover. In fact, inventory turnover is often not necessarily a linear relationship with other variables but is non-linear relationship. In this paper, we use an improved neural network. It can overcome the shortcomings of the traditional neural network that is lack of training samples to reduce its accuracy of calculation. Improved neural network firstly uses panel data model to extract the variables playing significant roles in the inventory turns and put them as inputs of improved neural network. The regression coefficients of significant variables of entity fixed-effect model are initial weights of improved neural network,using an improved ELM that improve the learning speed at the same time, simplify the structure of the model, and combining the data sets listed retail enterprises in China Simulation results show that the algorithm quickly and efficiently.