The Research On Computational Verb Decision Tree Classification Algorithm With Concept Similarity And Its Application On The Futures Market

As the development of the retail online transaction processing, financial market, sensor network and so on, there are a large number of data flows with rapidly changing and potential infinite, among which exists plenty of useful knowledge. Thus how to excavate out these unknown but valuable information from the data stream in order to guide people to make decision currently, is one of the hottest and most difficult problem that the current data mining field faced. Especially for the implied concept drifting data streams, how to research a classification method to adapt to the concept drifting data streams, is the focus of this article.Meanwhile, the futures market is a very complex nonlinear dynamic time series system, who transmits the price information represents the expectation of future supply and demand. Therefore, to forecast the price trend of the futures market, has extremely important significance for the development of national economy, government supervision of the market, the investors maximization the investment retained utility and so on. Based on all of these, this article puts forward computational verb decision tree classification algorithm within concept similarity. This algorithm not only introduced the concept of computational verb, which makes the decision tree have the dynamic forecast effect, but also stored the concept that appeared in the data flow, when the concept appeared once again, the algorithm searched and matched the historical concept through calculating the concept similarity, then using the corresponding classifier of one or several historical concept with high concept similarity to classify the current data. It greatly accelerates the speed of classification, which makes the algorithm more suitable for the real-time prediction of futures market.Firstly, this article describes the background of data mining and the related research on the quantitative investment in the futures market. It also pointed out that the strategies of the quantitative investment in the futures market can achieve better results after applied the data mining knowledge. Then gives a simple introduce to the basic theory and algorithm of data mining areas, and the existing methods quantitative investment in the futures market. Secondly, this article puts forward the decision tree classification algorithm implicit concept drift within concept similarity. The algorithm calculated the concept similarity between concepts through defining the concept, and the concept with high similarity can directly categorize the data using the historical concept classifier without training. Meanwhile, this article also introduces the computational verb decision tree algorithm. It through importing the concept of computational verb made the decision tree reflect the dynamic change process.Finally, this article put forward a new computational verb decision tree classification algorithm with concept drift with combining the above two algorithms. This algorithm contained the advantages of these two algorithms. Then the article proved the algorithm efficiency through the experiment demonstration analysis on the futures market.