Design And Implementation Of Labor Analysis And Forecasting System Based On Big Data Of Chain Sale

The retail chain has the features of many stores, many employees and so on. Labor costs rate is very high. All the retail chains are trying to raise their profit margins, by promotions, membership or cost control. In fact through effective control of labor demand, it can help to reduce the waste of human resources, and reduce labor costs at last, it is one of the effective ways of the cost control.Now it is a big data world, we are talking about the data mining in every walk of life. Retail chain is the powerful one obviously, because it is relatively early into the information age, and produced a lot of marketing data over the years, and the data is recorded and saved. The enterprise needs to use big sales data to make an in-depth exploration of profit increasing. On the other hand, the sales data has the characteristics of mass quantity, widely distributed, rapid increase. And it was apparently influenced by factors such as promotions, social policy, holidays, or important event. It increases the difficulty of future analysis prediction based on chain sales.For the type of forecasting on chain sales has a difficult and challenging, this paper deals with the design and implementation of labor analysis and forecasting system based on big data of chain sale. Design the data integration, enterprise labor models. With full consideration of various factors that have effect on sales, establishes forecast model based on data mining tools and algorithms.This paper describes the demand analysis, the outline design, detail design and application system implementation using an object-oriented formal method. Adopt the UML modeling tools to describe the design and implementation method for main modules and key business, using use case diagrams, sequence diagrams etc. This system selects JBoss Seam framework, a three-tiered B/S/S structure. We use Kettle to do the data integration. We forecast customer service metrics quantity by time series algorithms from Weka and other data mining tools. Use the stored procedures to improve the items forecast performance, in order to fix the data mining algorithms slow performance issue, we selected JPPF, it can parallelize computationally intensive tasks and execute them several node servers.This system has been deployed in Big Y Inc., an U.S. supermarket chain. The system running stability and the functions required. They forecast sale items, non-sale items and metrics well by this system, they created labor model for all stores, help them forecast labor demand come true.