You are in the process of configuring a k-fold cross-validation for a classification task, where 'k' represents the number of data splits. Given the necessity to balance between computational efficiency and a comprehensive evaluation on a limited data sample, determine the most appropriate value for 'k'.