The Nationwide Retailers’ Data Processing And Market Awareness Based On Hadoop

With the development of the Internet in recent years, contributing to the "Internet +" Rise, tobacco companies already have a huge Retailers’ information data. Number Retailers in the country has more than 8 million, will produce a large number of orders per month information, over time, the order information data has reached the level of TB. However, business was originally set up hardware and software does not have to store and process the data capacity, thus leading to valuable information that can not be extracted to form an awkward situation. In this paper, the country continues to surge Retailers information data with existing large data storage technology, based on the architecture proposed Nationwide Retailers orders Hadoop data analysis system, a detailed analysis of the Hadoop cluster node storage performance and response time, and a massive storage and processing of data. ARIMA forecasting model based on the establishment of a sales forecasting model, and the model is validated, according to forecasts based on the forecast proposed marketing model, also presented to Retailers lifetime value calculation, provide decision-makers to Retailers loyalty degree promotion strategies. Specific contents are as follows:1) According to the nature of the nationwide retailers’ order data, a data storage model based on Hive is proposed, which is used to partition the massive data. Using the model to solve the problem of the application of the nationwide retailer order data storage and load balancing.2) Discussion of the entire HDFS common data processing methods, and storing data according to the actual situation, proposed an algorithm to reduce data processing in heterogeneous cluster response time, the algorithm is to analyze the nature of the cluster and the cluster of processing data execution response time is proposed based on the performance data computing node allocation strategy. Retailer for national data, using the allocation policy to establish a data processing model, reducing data transmission on the network to prevent data congestion, provide cluster timeliness.3) Enterprises with keen market awareness, the ability to grasp the market dynamics and direction of the market, the key lies in big data mining. From the perspective of market awareness and market responsiveness, in a large data base on the use of ARIMA forecasting model to predict sales marketing, and Retailer put forward a method to calculate the market value, loyalty, important decisions to ease reliance on personal experience for the brand launch and market response provide effective information support.4) Data storage technology based on Hadoop established a large data processing platform, design a to produce enterprise retail data traction for the ecological cycle of oriented data storage, and the strategy of data platform in circulation and in data and information as the core of the data synchronization of internal ecological cycle of three ecological circulatory system architecture.