We examine the feasibility of Deutsch-Jozsa Algorithm, a basic quantum algorithm, on a machine learning-based logistic regression problem. Its major property to distinguish the function type with an exponential speedup can help identify the feature unsuitability much more quickly. Although strict conditions and restrictions to abide exist, we reconfirm the quantum superiority in many aspects of modern computing.