中文摘要：

提出基于修改的代价敏感学习的方法对不平衡的天气数据进行预处理,结合天气数据自身的特点,以单位时间的降雨量为成本的值,将数据合理有效地区分为下雨和非下雨两类;进而运用基于逻辑的方法对处理完的数据进行分析,运用分支限界算法得出布尔分类器。实验结果表明此方法可行有效,该方法可进一步对布尔分类器结果进行逻辑运算,从而达到更加灵活的操作分类器的效果。

英文摘要：

This paper proposed the modified cost-sensitive learning methods to preprocess the imbalance weather data. Considering about the specialty of weather data,it made the value of rainfall per unit time as the cost value. So the data could be divided into two types of rain and non-rain effectively and reasonably. And then it used logic-based approach to analysie the data processed,used branch-and-bound approach to derive a Boolean classifier. Experimental results show that this method is feasible and effective. What＇s more,it＇s valid to perform any further logic calculation or logic operation on the result of the Boolean classifiers,achieving more flexibility.