Research On Analytical Methods Of Agricultural Produce’s Market Price Based On Web

In recent years, the abnormal fluctuation of market price of produce has become a focus of attention in our society, which not only has a direct influence on farmers’income and their enthusiasm for production, but also plays an important role in common people’s daily life and their vital interests. A macro-management of prices is especially important to ensure a stable and healthy development of economy and to secure people’s lives. The prediction and monitoring of price is an important element to maintain a stable price, at the same time a precise monitoring as well as a prediction with low error rate is an important part for agricultural research. It is urgent to establish a vertical search engine system to regularly record produce’s prices on the Web because the information there is usually widely distributed and quickly updated in our country. A platform to analyze, monitor and predict produce’s price is imminently needed so as to provide comprehensive and clear results for the reference of governmental production regulation and decision-making, as well as farmers’ planting, and to make a contribution for the stability of produce’s market price.The information released by some agricultural website has the problem of non-standard product name and non-uniform data unit. Through the analysis of market price of various produce from different websites, the research puts forward the following7principles, such as the standardization of produce name, that of market and data unit, the initialization of produce categories and provinces, the choice of a material-removal strategy and data processing of a price around zero, studies the tree system of DOM, the regular expression, extracts text messages from the Web through HTMLParser, establishes a vertical search engine system by taking advantage of softwares such as Heritrix, and finally forms a unified and integrated price-database of produce in the form of SQL server.In order to improve the prediction accuracy of produce’s market price and timely find the produce with an abnormal market price, the research selects carrots, white turnips, Chinese cabbages, green onions, kidney beans, cucumbers, Chinese chives, eggplants, peppers, green peppers, potatoes, tomatoes and rapes to make a comparative analysis with prediction algorithms. Meanwhile, the methods to set up5kinds of weights are comparatively analyzed with the weighted arithmetic average prediction, the experiment results show that the latter is better than the other five methods if current year’s prices are taken as with the lowest errors of weights, and the prediction with mean value is superior to a simple arithmetic average method. In addition,11different kinds of time-series non-seasonal exponential prediction methods are also studied, the results show that the curvilinear trend duration method of B-Gompartz and double exponential are not suitable for the prediction of produce’s market price, among the methods of simple and weighted arithmetic averaging, of series with an average increment and with geometric averaging, of single and double moving averaging, of single and double exponential smoothing and of a rectilinear trend duration, the double moving averaging and the double exponential smoothing are superior to other methods and suitable for the prediction of produce’s market price for their lower errors. On the basis of the experiment results, the research puts forward an improved prediction method of double exponential smoothing which is superior to the previous one and the double moving averaging for its lowest errors, when there is a different quotient of single and double exponential, all the sums of squared errors is no more than those when there is the same quotient. At the same time, the research compares the horizontal prediction and the tendency prediction with seasonal indices, the results show that there is a distinct difference between the sums of squared errors with these two methods, and the actual predictive price should be predicted with the lower squared errors. As to the determination of produce with an abnormal price, the research puts forward two methods, that is, the squared errors of predictive price and actual errors as well as historical errors, so as to determine the produce with monthly abnormal price and that with an abnormal price last year and this year.Websites which provide produce prices usually give an initial price. On the basis of data taken from various websites, the research develops a Web-based analytic system of produce price with the help of MyEclipse enterprise workbench so as to achieve the function of price search, price analysis and price prediction, and so forth. The function of price analysis includes the analysis of price trend, the comparison of price from different provinces, that of different varieties, that of chain relative and year-on-year ratio and that from different markets. The function of price prediction includes the prediction of unit value and that of trend, and the comparison of price from different provinces includes the comparison of price of a certain day and within a certain period of time, and so on, meanwhile the analytic results are shown in the form of line charts, bar charts or maps with a beautiful interface and a practical function. The intelligent Web information analysis method of market price of various produce can help the agricultural administrations, agricultural enterprises and farmers master the variety of price and the tendency of different regions and different produce.