This master's thesis is concerned with establishing a lower complexity bound for solving large-scale binary logistic regression problems via first-order methods.