An application of Sigmoid and Double-Sigmoid functions for dynamic policyholder behaviour