|
Artificial intelligence-based decision support technology is an important development direction of modern decision support, and the rapid development of the neural network technology to intelligent decision support opens up a promising development direction. 1980s Inmon proposed the idea of ??a data warehouse, the huge success, traditional data mining techniques, so he is facing new problems and challenges. Vast amounts of data stored in the data warehouse, data mining researchers start from a fresh angle to research database knowledge discovery (KDD) and other issues. Here we design and realize a prototype of a data mining model based on artificial neural network with fuzzy output, this model is actually an intelligent decision support system, capable of processing the raw data to users, the use of already obtained knowledge of user data to judge the results of the system decision and explain to the user in an appropriate manner. The model we achieve an organic data processing and decision-making system, the first model to accept the user usually MIS system has accumulated a variety of data, the original \their pretreatment, this process is actually quite tricky, the most typical, it is first of all to integrate data from different data sources, the Secondly, it is to be able to be cleaned to such data, remove dirty data that we are not interested, Finally, it is also a re-interpretation of the data, processing data standardization, consistency and other issues. Necessary data preprocessing, we can make decisions on this data classification. The model we realized, this is the core of the entire model, our model using artificial neural network technology as the basic data mining model with fuzzy output. In the basic model of the network, the use of standard data provided by the user to train the network, making it remember the standard data model is the key of the entire network of intelligent decision. In our model, on the one hand, to take into account the question of the ability of the network correctly classified, on the other hand, we also have to consider the efficiency of the network. Research and solve these problems in this model one of the main focal points, we will focus on the typical characteristics of the face of the user data in our model, the network application testing and transformation, so that the network can be for our data optimum efficiency and performance. Our model is implemented using the Java language, on the one hand, taking into account its adaptation of the model may be very beneficial, on the other hand, we intend to increase the security features in the model, and Java in this regard to do quite well, It provides models, algorithms, and standard class, we can easily Engineering, Beijing University master's degree thesis of our model system for the safe handling. It should be noted that our model integrated decision-making process of the user's data, however, this is just a model, it is not a product, and therefore, it is not yet mature. For example, it is dealing with the user interface, to do a very rough, this is because some of the inherent limitations of the Java language itself, we believe, should be the core of our model on the understanding and improvement of the network itself, The user interface is not our focus. However, even if the network itself, we have made the work is not perfect, as our understanding in this area is far from enough, is not well on the latest development of the neural network technology to capture and understand, and thus our network improvements are not necessarily the best. These are the direction we should continue our efforts.
|