Feature engineering plays a vital role in selecting relevant features that have maximum predictive value. In this paper, we have proposed a Gaussian fuzzy score function to rank the features in descending order of their score values. The mean and variance of the Gaussian fuzzy score function are determined using mean of k-middle. The mean of k-middle plays an important role to determine the complementary information of the features in the dataset. The features selected using proposed feature ranking method are fed to four widely used classifiers, i.e., linear kernel support vector machine, radial basis function kernel support vector machine, random forest and softmax classifier respectively. To show the effectiveness of the proposed approach, we compared its performance with that of state-ofthe-art methods on five large-scale gene expression datasets.