Abstract: As increase in data dimensionality classification of data increased. In industries or organizations fault detection is important task. Due to imbalanced of data classification process has problem. In standard algorithm of classification majority classes have priority for classification and minority classes have less priority for classification therefore it is not suitable for minority classes fault detection from data is applied for only majority classes and less for minority classes. Incremental clustering algorithm solved this problem but it reduced data attribute. To maximize the accuracy, time, and memory for this we proposed a feature selection algorithm for better performance of classification and fault detection.

Keywords: Classification, Class imbalanced data, Clustering, Data mining