Analysis of Machine Learning Classifiers for the Detection of Lung Cancer from Micro Array Gene Data

Microarray gene data is a good resource for studying gene expression patterns of various biological processes and diseases. The paper adopts a classification method of high-dimensional microarray gene data for lung cancer by utilizing dimensionality reduction and feature selection techniques. The dimensionality reduction is done by Short Time Fourier Transform (STFT) and utilizes Scalogram as the feature selection technique. The classifiers, namely Non-Linear Regression (NLR), Gaussian Mixture Model (GMM), Softmax Discriminant Classifier (SDC), SVM (polynomial) and SVM (RBF) are employed for classification. The classifiers’ performances are analyzed and compared. The SVM (RBF) classifier with the Short-Time Fourier Transform Dimensionality Reduction method and Scalogram 1024 feature selection achieved the highest accuracy of 96.13 % compared to other classifiers.