Application Research On Enterprise Marketing Decision Support System

With the rapid development of information technology,enterprise to improve its core competitive power in the market,enterprise requires constantly make scientific and rational decision based on accumulated huge data.Therefore,a decision support system that can guide the business manager systems is especially important for scientific decisions.Aiming at the present situation of decision support system at home and abroad,this paper studies the two core contents of sales forecasting and customer segmentation,the main work of this paper can be summarized as follows:1)Sales forecast.Prediction refers to itself on the basis of available information,according to a certain measuring methods and rules on the future.Accurate sales projections not only for scientific decision-making of enterprises,but also help enterprises to maximize profits.Paper first presents a selection of influential factors in product sales as the BP neural network model of input parameters;then BP model to determine the hidden layer node number,problems such as easily into a local extremum,constructed a genetic algorithm optimized BP network predictive model;finally,put this article through the construction of predictive models and comparing unoptimized BP models and predictive analytics proves BP neural network in optimization of genetic algorithm to obtain a higher predictive accuracy.2)Customer segmentation.Customer segmentation to enable enterprises to better grasp the customer base,provides decision support for precision marketing companies,stable core customers,attract new customers and maximize business efficiency.Based on customer loyalty customer segmentation and customer value of two dimension,build multi-dimensional cross-customer segmentation model.First of all,customer loyalty,in accordance with decision tree method "last purchase","shopping" and "frequency of purchase",losing customers,floating customers,and loyal customers;then,in view of customer value presented in affinity propagation algorithms k-means cluster analysis algorithm to customer,customer is divided into low-value customers,average customer,focus on the development of customers with high-value customers;finally,to cross the two dimensions more accurate customer segmentation models,and for each type of customer segments with appropriate marketing strategy for policy makers.