Microarray gene expression data sets are huge in number. Hence, we have devised a procedure to optimize and simplify such multi-dimensional data to represent it in lower dimensions. Furthermore, we also have demonstrated an important test to determine whether a test sample of genes is cancerous or not. By mapping the test sample with a few cancer-affected gene samples, we have grouped the optimized data set to check the number of clusters that will help us to identify the affected genes. Therefore, this work reports an easy optimization technique to identify the cancer-affected genes. Kohonen’s self-organizing map along with entropy, symmetrical uncertainty has been employed to develop the software tool to detect the cancer-affected gene in a fuzzy framework. The proposed work performs best with respect to some existing models, e.g., ANN, SVM-RFE, Apriori, and FP growth in predicting the true positives.