Within GLMs framework, the mean of the dependent variable is related to the linear
predictor through the so called link function. It is well known fact from the literature that a
logarithmic function is the natural link function for the Poisson distribution: