Data To Predict Based On BP Neural Network And Report Generation Technology Research

With the rapid development of computer networks, in the information society of today, more and more data, has ushered in the era of big data. It’s very important for businesses to manage large amounts of data effectively. Report is an important tool as an information organization, statistics, analysis of report, it has an extremely important role in enterprise operations management process. Currently there are a lot of active reporting system on the market, foreign reports have Crystal Reports, Active Report, Style Report, Windward Report and so on, more popular domestic reports have run dry, UF table(cell), Fine Report, e lists. But companies using third-party reporting tools not only spend the high cost, and corporate data security can not be guaranteed. For the ever-changing needs of enterprises, using the third-party reporting tools can not be timely response. Moreover, long existing reporting tool develop long cycle, low efficiency, requires a lot of developers. It’s more difficult to learn.Configuration process need to write code. In summary, the flexibility of the domestic reporting tools can not solve the enterprise applications that exist, versatility,timeliness, portability and other issues.To solve the above problem, this paper proposes data to predict based on BP neural network and report generation technology research. The main work is as follows:Firstly, the existing prediction model algorithm to compare the advantages and disadvantages of each model, to identify the most appropriate data prediction model of the automated reports. In this system, the eventual adoption of BP neural network to forecast data. Depending on the data to predict demand, structural design appropriate BP neural network, through trial and error to determine the final three-tier network architecture. You can determine the input layer and output layer neuron number, depending on the report instances, and then to determine the number of neurons in the hidden layer by experiment to try and experience. Transfer function of hidden layer using a logarithmic function-S shape, the training function using LM algorithm, based on the principle of decreasing the error, by continuously adjusting the network weights and thresholds, so that the final results to within a preset error range of accuracy. Experimental results show that the preferred neural network design properties can be adapted to the actual demand.Secondly, the automated reporting system in the front page design proposes anew idea, the front page abstract condition, tables and graphics in three parts. Use jqGrid Spreadsheet Add-form design, to operate the page display and tabular data.Graphic design use Echarts chart widget, Echarts provides a very rich graphical structure, meet the needs of the company’s graphical reports and supports mashup map and graph. MVC model using three-tier architecture design in the architecture,design from the presentation layer, business layer and data layer three. The automated report short development cycle, the actual realization of the company’s reporting needs, you can automate the configuration, without writing code, it frees developers.