This document presents an empirical analysis of the Fitness-based Area-Under-Curve - Bandit (F-AUC-Bandit), an adaptive strategy (or operator) selection method recently proposed in the context of Genetic Algorithms. It is here used to select, while solving the problem, the strategy to be applied for the next offspring generation based on the recent known performance of each of the available ones, within a Differential Evolution algorithm applied to contin- uous optimization problems. Experimental results are obtained on a testbed of single-objective noiseless functions. The performance gain achieved by the use of adaptive strategy selection methods is shown by comparing F-AUC-Bandit with what would be the common naïve choices: the use of a single strategy or the uniform selection between a sub-set of available strategies. F-AUC-Bandit is also compared to previously proposed adaptive schemes, showing a significantly better performance (w.r.t. expected running time to achieve a target solution) on most of the functions, while presenting a robust hyper-parameter setting. Although still being not competitive with state-of-the-art continuous optimizers such as the CMA-ES (to which an empirical comparison is also presented), a big enhancement is achieved over the use of the basic Differential Evolution, while also improving over both naïve and existent adaptive methods.