Fuzzy case-based reasoning: Weather prediction

Classical K-nn algorithm in CBR has been used widely. But in the actual situation, cases often have kinds of features, and classical K-nn algorithm cannot tackle it well. Other fuzzy K-nn algorithm may apply well to these perspective systems, but they do not adapt to weather prediction. In the paper, we propose our novel fuzzy K-nn algorithm. Because weather is continuous, dynamic, and chaotic, in our algorithm, the time function as the adjustive factors is introduced to similarity-measuring function. Fuzzy logic is used in the retrieval of cases. Then fuzzy K-nn algorithm is proposed. Finally, Experimental results are given and we get better effect.