Sporadic Motor Neuron Diseases (sMND) are a group of neurodegenerative conditions. It causes severe damage to the nerves in the brain and spine, makes it lose the function over time. The progression of this disease has a strong relationship with the genetics of an affected individual. Analyzing the gene expressions of sMND affected cases unveil the diagnostic genetic markers of the condition. But, higher dimensionality of the data affects the predictive performance due to the presence of vague, imprecise features. To address these issues, an effective hybrid feature selection technique called Correlation-based Feature Selection-L2 Regularization (CBFS-L2) is proposed to identify the candidate genes of sMND by eliminating inconsistent, redundant features. The proposed CBFS-L2 model revealed 26 significant Single Nucleotide Polymorphism (SNP) gene biomarkers of sMND. The performance of the identified subset is evaluated with four state-of-the-art supervised machine learning classifiers. The proposed feature selection technique attained a high accuracy of 94.31% on sMND dataset, outperformed benchmarked results, and other feature selection techniques. © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd 2021.