Introduced on day 3
Overdispersion
Count data whose variance exceeds the mean, more spread than a Poisson model allows.
In detail
Usually caused by rates that vary between units or by clustered events (group bookings that no-show together). It fattens the tail; the negative binomial distribution is a common model for it.