In this thesis we present two parts: Adaptive storage rearrangement, and process discovery, where the second part adds to the work we did in the previous semester. The work for this thesis has been carried out in collaboration with two stakeholders, Av Form which is a warehouse located in Denmark, and Logimatic whom develop the warehouse management system called LOGIA, which is being used at Av Form.

In the First part of the thesis we present the warehouse and how part of their work is structured, followed by the specification and clarification of an area which a worker at the warehouse believes could be improved. We come to the conclusion that the area suited for optimisation relates to the process of order picking, which is believed by many to be the most expensive part of running a warehouse.

We proceed to explain how work done through LOGIA affect the configuration of the warehouse, i.e. where items are stored, their quantity etc., and how this work is being logged in LOGIA. We then present our theory which on the basis of the LOGIA log, is used to devise a novel algorithm that quantifies the impact of rearranging storage locations for a given warehouse configuration. Using this algorithm it is possible to predict which rearrangements in the current warehouse configuration will result in reduced costs for the warehouse.

We then include time heuristics of specific operations used as part of the prediction, before we introduce our implementation of the rearrangement algorithm in the LOGIA system---an implementation which have been used by Av Form.

The LOGIA implementation concludes the first part of our thesis, before we proceed to the second part regarding process discovery. Based on work we conducted during our previous semester, where we amongst other work presented a method for the discovery of timed-arc Petri nets, we elaborate on the time dilemma which emerges during the discovery process. To solve this issue we propose two methods which preprocess a timed event log before it is used in the discovery process.

After our elaboration of the time dilemma and its solutions, we re-introduce our method for calculating the distance between an extended timed-arc work net and a discrete-time event log, where we include an example based on the context of our current stakeholders.

Our thesis then culminates in four experiments covering different areas of the work we have conducted. The first experiment serves as an evaluation of our rearrangement algorithm, the second experiment is used to evaluate the solutions to the time dilemma, our third experiment covers our data mining and process discovery efforts, where we use the solution for the time dilemma chosen in the previous experiment. In our fourth and last experiment, we see how our distance measure evaluates on the discovered models and discrete-time event logs from Av Form.

And finally we conclude on the entire thesis and present suggestions and thoughts on areas that can be researched further in the future.