Font Size: a A A

The Application Of Data Mining Technology In Agricultural Network

Posted on:2013-07-08Degree:MasterType:Thesis
Country:ChinaCandidate:Z X ZhouFull Text:PDF
GTID:2208330434970258Subject:Computer technology
Abstract/Summary:PDF Full Text Request
In order to better guide data stream mining of the agricultural internet of Things, we first analyze and summarize the characteristics of the data on agricultural internet of Things, and write a data mining requirements document. We developed the SVM program, which has high accuracy, to implement a demonstration system and also use it in the classification of non-stream data. We used SVM program to classify the sample data, marked them with normal and abnormal symbol, and shown the mining results in a web site, so that the results can be more vivid and easy to be understood.Secondly, in order to mine multidimensional data stream more accurately, we developed a data integration platform. There are a large number of single dimension data stream in agricultural IOT, but lots of them are interrelated. So, it will need to integrate these single dimension data stream into multi-dimensional data stream. Our data-integration platform can call different integration programs to transfer the data stream into multi-dimensional data stream depending on its type.Then, we developed data stream mining algorithms adapting for agricultural IOT based on the analysis of the data characteristics of the agricultural IOT and referred to the existing state-of-the-art data mining algorithms. We improved CVFDT classification algorithm and made it not just can process stream data efficiently and can adapt to concept drift, but also can deal with numeric data which is common in agricultural IOT, and can choose the best parameters to increase the accuracy.Finally, we analyzed how to predict the shelf life of agricultural products. With the comprehensive data information in aspects of agricultural product growth, processing, storage, transporting and sale, we can predict and evaluate the shelf life of agricultural products to better guide management and marketing according to different quality of agricultural products. So fai, we have completed the requirements analysis, the data format definition, algorithm selection and some other key issues.
Keywords/Search Tags:agricultural internet of things, data mining, data stream, data integration, predicting
PDF Full Text Request
Related items