Návrh a implementace metody SANE pro realistickou interpolaci řídkých dat

This thesis deals with the reconstruction of continuous process intensity from interval-aggregated data. The work is motivated by a practical capacity planning problem in airport passport control, where only estimated passenger counts in fixed time intervals are available. The thesis proposes the SANE method, Smooth Area-preserving Non-negativity Estimator, which constructs a continuous intensity curve while preserving the volume in each interval, maintaining non-negativity, and ensuring smooth connections between neighbouring intervals. The method uses a local classification of intervals based on the relationships between neighbouring volumes and constructs corresponding analytical functions for different types of local behaviour. The proposed approach also includes special constructions and a propagation mechanism for cases in which the basic local shape cannot be directly realised. The proposed method is implemented as a computational prototype and subsequently empirically validated on operational data. The validation assesses not only the accuracy of the intensity reconstruction, but also the impact of the resulting curve on subsequent queue modelling. The results show that SANE can transform limited aggregated inputs into plausible continuous profiles that can be used as a basis for operational evaluation.