Understanding temperature-rain data using ID3 based concept reduction technique in FCA

Proper understanding of rain yield along with the relevance factors and their extent of relation in the yield of rain is very important to maintain a smooth life style in every one's life. The definitive classification of mean temperature and heaviest rainy days depends on the weather changes that occur seasonwise during any year. In reality, visualizing the effects of climatic changes such as temperature in the rain-yield during over a period of years is very difficult and there is no method or tool to help us in this aspect. Formal concept analysis (FCA) which is a mathematical model that expresses the relationship between various features and entities in terms of pairs called concepts. These concepts are hierarchically related to form a unique concept lattice which is a diagrammatical view of the information available. In this paper, an approach to facilitate the understanding of temperature-rain data of Vellore district with the use of data collected for the recent 15 years period is presented. For the analysis, the mean temperature data is preprocessed seasonwise over the years. In a similar manner the rainy days also preprocessed seasonwise over the years. We illustrate the process of extracting meaningful information from the data with the use of FCA. In this process an ID3 algorithm based method is employed to identify more important features from the context. These important features are used to compress the concepts obtained from FCA and thereby reduce meaningful information. The efficiency of the proposed method is validated using few efficient metrics available in literature.