Most ‘predictions’ of the rate of meteors per hour during meteor showers are based on both theory and observation. Essentially, a computer model is built containing the trajectories of every known comet – since it is the debris from comets that forms the ‘stream’ of particles we see during a meteor shower.
This model contains information on the rate that these comets release material, along with the sizes, directions and velocities at which they are released, as well as the gravitational forces that determine their subsequent trajectories through space. The trajectory of the Earth and the conditions of the Earth’s atmosphere are also inputted into the computer model.
By watching how Earth moves through the meteor stream it is possible to estimate the likely number of meteors that will be visible during a given shower for a given location. But different astronomers use different models. Plus, these models are partly based on difficult measurements of the meteoric particles in the Solar System, so their predictions are often only approximate. But generally, they can be used to reliably predict when a meteor shower is likely to be more or less intense than the average.