E-commerce Customer Behavior And Sales Forecasting Model And Realization

In the e-commerce environment,the customer's access to e-commerce web site and behavior of purchase transactions would cause large amounts of data.These data contains a great deal of potential information for analyzing and forecasting market. Analyzing these data in depth can not only provides decision-making basis for enterprise to improve e-commerce web site's quality,but also has important practical meaning of increasing e-commerce's operational efficiency.This paper chooses web log data and historical data of customer purchasing as the basis for predicting client behavior,uses neural network and markov chain measure as modeling tool,researches the e-commerce problems such as client's access behavior and commodity's forecasting sales.The major work of this paper is as follows:Firstly,the model which bases on markov chain to predict the e-commerce client behavior is proposed.Based on web data log,uses markov chain model as modeling tool,establishes a prediction model of e-commerce client access modes of transfer, projects a measure for computing transition matrix from Web log data and the concrete predictor step,and verifies this model's feasibility and validity from some electronic store's client access data.Secondly,the model which bases on neural network to predict the e-commerce merchandise sale is proposed.Because there are many factors which can effect on the merchandise sales,the client's purchasing behavior is an intense non-linear process. So using conventional prediction model to forecast sales has some limitations. Therefore,this paper establishes an e-commerce's merchandise sales prediction model by using neural network model,designs the definition of net hidden layer and learning algorithm,and verifies this model's feasibility and validity from some online bookstore's sales data.Thirdly,the prediction model's prototype system above-mentioned was designed and implemented.Designs and implements the prototype system of e-commerce client behavior prediction model and merchandise model by using J2EE + SQL server technique,analyses and tests by experiment.Experiment shows that this paper's prediction model is correct and feasible.The outcome research of this paper provides a degree of reference for e-commerce operator to predict client behavior and merchandise sales.