Research For Effective And Sales Forecasting Of E-business Promotion Activity Based On Data Mining

In recent years,with the rapid development of the Internet and e-business,it is more competitive among enterprises,so e-business platform and enterprises will use a variety of sales operations methods to seize the market.In order to promote sales,the enterprise will participate in various types of promotional activities,which are provided and organized by the e-business paltform.Meanwhile,with the rapid development of information construction,enterprises have accumulated a lot of historical data.Therefore,the paper is based on the method of data mining and data analysising,through historical data for e-business activities effect analysising and activity sales forecasting.Through data mining and analysis,it is to enhance the scientific degree and intelligent degree of enterprise's management decision-making,so that this research has both theoretical significance and practical significance.Firstly,the paper analyzes the short-term effect of e-business promotion activities,mainly focusing on the two aspects of sales and passenger traffic,and the effect of sales and passenger traffic both before and during the promotion activity.Based on the historical data of a tianmao's flagship store,through statistical analysis of t-test and regression analysis,it is obvious that in terms of the activity effect of sales,sales during the activity period will increase significantly and sales before and after the activity will decrease.Besises,about the activity effect of passenger traffic,the traffic during the acitvity will increased significantly,too;while the traffic before and after the activity will be no differences.Decrease in conversion rate leads to lower sales in this period.Secondly,according to the decision-making problem of whether or not the enterprise should participate in an activity,combining with the above analysis,the paper makes a decision-making model based on overall profit,and comprehensively considers the income of the participating and non-participating goods during the period of the activity and the period before and after activity,as well as commission fees charged by the e-business platform.Through the analysising of decision variables,it is found that the change of sales during the daily period and before and after the activity is relatively stable,and can be easily determined by the method of moving average.But the sales during the activities period has large changes relatively,the paper proposed an integrated forecasting model,which based on support vector regression,combined with particle swarm optimization and gray comprehensive analysis,for sales forecasting of participating in the activities;used association rules to analyze the interaction between non-participating goods and participating goods.Thirdly,the paper is to describe the data preparation process,including data source analysis,data warehouse design and data ETL implementation.Then combining with the method of the paper to do the example analysis,it is concluded that the integrated forecasting model has an improvement on the prediction accuracy compared with the single support vector regression model,meanwhile,combining the interest degree can find more effective association rules,besides,taking into account the loss of profit before and after the promotion activity of the decision-making model is more comprehensive and objective,and helping enterprises making better management decisions.Finally,concluding and evaluating the research,besides proposing the further reaearch.