Development of the Data Preprocessing Agent's Knowledge for Data Mining Using Rough Set Theory

摘要

Data preprocessing is one of the important task in Knowledge Discovery in Databases or Data Mining. The preprocessing is complex and tedious task especially involving large dataset. It is crucial for a data miner to be able to determine the appropriate data preprocessing techniques for a particular data set as it will save the processing time and retain the quality of the data for data mining. Current data mining researchers use agent as a tool to assist data mining process. However, very few researches focus on using agent in the data preprocessing. Applying agents with autonomous, flexible and intelligence reduced the cost of having a quality, precise and updated data or knowledge. The most important part of having an agent to perform data mining task particularly data preprocessing is the generation of agent's knowledge. The data preprocessing agent's knowledge are meant for agent to decide the appropriate data preprocessing technique to be used on a particular dataset. Therefore, in this paper we propose a methodology for creating the data preprocessing agent's knowledge by using rough set theory. The experimental results showed that the agent's knowledge generated is significant to be used for automated data preprocessing techniques selection.