The transportation of goods and people is a key element of the economy. With the ever increasing intensity of road traffic, several issues arise that need to be addressed. Policy makers require reliable forecasts of traffic behaviour in order to develop an effective policy. Activity-based models are best suited to make the predictions used for these policies. This master thesis will investigate the performance of one such activity-based transportation model, FEATHERS, when presented with decreasing amounts of training data. This study will attempt to do this by methods of progressive sampling, evaluating both the value of parameters derived from the output and the performance of the decision trees, which form the fundaments of the decision process in the activity-based model. The goal is to be able to report on a minimum amount of training data required for the model to function at an acceptable level. Chapter one forms the introduction to this master thesis. The background of the r