First, you can find if the method is reliable for the sttudy of properties of your molecule (or related) in the literature. In addition, test computations in some molecular properties such as vibrational frequencies, UV-absorption, can help you with the selection of the m,ethod, expecting a low deviation with respect to empirical parameters if available.
Not a good choice. APFD is not a popular functional, and it doesn't exhibit good performance (see e.g. Phys. Chem. Chem. Phys., 19, 32184 (2017)). B3LYP-D3(BJ)/6-311G** is not only popular but also robust. Many papers showed that if frequency correction factor is employed, B3LYP is still one of the best functionals for evaluating vibrational frequencies, at least for organic systems (see e.g. J. Chem. Theory Comput. 2010, 6, 2872–2887)
Its quite a small system. You can quickly try yourself and compare to experiments.
APFD is the functional of choice in the book of the Gaussian16 website (http://gaussian.com/expchem3/). I do not see why not to recommend for such a simple system.