For every given day, it is known which percentage of guests fell into each of the following age categories: 0-5, 6-14, 15-20, 21-30, 31-60, 61-80, 81-100.Age information was not available at the level of the guest, so it was not possible to identify individual guests. A descriptive analysis of age group distribution per day shows that the guest composition by age on a specific day falls into one of two age segments where one segment has a higher proportion of younger and the second one a higher proportion of older guests. We use these groups instead of the original age group distributions to minimize problems that may arise from multicollinearity in the data.