The scheduling strategies used at blood collection centers worldwide vary considerably. At the Blood Bank of Oslo, fixed appointments are used. However, on average, only 60% of blood donors with an appointment actually show up. Overbooking is therefore necesarry, but because the day-to-day variations in donor arrival are large, this may lead to long waiting times before donation. Keeping waiting time short is important to motivate for repeat donations, which is crucial because of scarcity of blood donors. Optimal overbooking must satisfy the need for enough donors each day, and on the other hand keep waiting time short, to ensure sufficient numbers of donors in the long run. Due to the day-to-day variations in arrival, the optimal number of appointments to make varies as well.

If one could predict the number of no-shows the day before the appointments, one could compensate for these by making extra appointments at a short notice. Such predictions are possible because the probability to show up varies systematically with individual characteristic such as age, gender, how long the person has beeen a donor etcetera, in addition to vary by day of week and day of year. The goal of this project was to build a statistical model for predicting the number of no-shows a given day, based on information on each donor with appointment that day. The model will be implemented at the Blood Bank of Oslo in 2005.