A novel k-means clustering approach using two distance measures for Gaussian data