One method would be to take data from ticket sales at a previous concert where a similar number of tickets were available. Analyse these statistically to find the average number of tickets, the mean, median and standard deviation of the dataset. You may need a random number generator scaled to the range of the lowest to the highest number of tickets sold. The device will need to monitor the simulated sales so as to maintain the basic statistical parameters: mean, median, standard deviation. If a random number generator is unavailable or undesirable you can scramble the data so as to deliver the individual number of tickets in a different sequence to the original. This means that the statistical parameters will not need to be monitored because the dataset hasn't been changed, just the order in which the tickets were originally sold.