Research On Date Mining Model Of Grain Marketing Information System Based On Rough Set

It should be said that there has been great improvement in respect of data collection and storage in last decades. However, the amount of real-life information is increasing exponentially. We are lack of ability of understanding and digesting the great deal of the data. So there is a urgent requirement of new tools and technologies. Many researchers in other domains are stimulated by KDD(knowledge discovery in databases), and have many interests to study KDD. In a lot of fields, it is RS(rough set) that have made prominent contribution to the KDD. Data Mining methods based on RS are efficient ways between data reduction and data mining. RS by Pawlak has become a well-known method of KDD. One distinct feature of RS is that in the original model, it mines information only by data, while most statistical and machine learning methods need hypothesis and parameters of the model. Another feature of RS is that it can deal with fuzzy and uncertain knowledge. It can deal with not only decision systems but also information systems. In our paper, the reduction of data based on RS is shown in two aspects: attributes reduction and values reduction. Attributes reduction is to seek the minimal subset of condition attributes to substitute the whole condition attributes, on the premise of that it does not miss the semantic information as concerning decision attributes. So we realize the transverse reduction. Values reduction is actually based on attributes reducts. In allusion to each regulation values reduction try to generalize it to make it more brief and representative. It presents not only the original method to seek the reducts but also the methods based on discernibility matrixs and heuristic algorithms. With the developing of RS, many variations and improvements of RS have been presented, such as VPRS(variable precision rough set model) etc. In our paper, only a little thinking is introduced but not the detailed. The main idea of our RS research is how to implement the data mining by data reduction. Each step such as data preparation, data preprocessing rules creation and decision algorithm related to is only summarized. Our aim is to apply the data mining method based on RS to the practice. Aim to the management information of crop market and the actuality that intelligent technology research of national crop market information analysis is still underway, RS is applied to the crop market and analysis of crop market information. This paper makes the experiment for the first time that applies RS to the management information of crop market, and obtain a good effect. This paper's main production is that a data mining model system is designed and implemented based on RS. The system consists of data preparation, data preprocessing rules creation and decision algorithm. Potential and valuable rules are discovered by the system in the historical data of crop management futures and merchandises on hand. The system is used to direct decision of crop management, so that management and finance risk can be reduced. It provides technology support to the decision processing of crop management market.