This thesis investigates methods for the analysis of log files from warehousing systems with the aim of evaluating performance. For the practical problem at hand, the data originate from different warehousing systems where human-machine interaction plays a major role. More precisely the log files are created by warehouse control systems in response to human and machine activity. The aim of this thesis is to determine the temporal activity of humans in the warehouses by analysing the logs for the purpose of productivity measurement. To accomplish this, three different approaches have been developed: One approach incorporates additional a-priori knowledge in the form of expert knowledge; the second assumes that there is a characteristic temporal distance between two logs for the same activity; the third approach uses entropy as a measure of information content, in this special case information content as a function of time - assuming that entropy is increasing during working times and stagnating during pauses. Histograms and thresholding are key concepts which have been used in each approach to identify the working time. The three approaches have been evaluated using ten days of log data from two different warehouses with different degrees of automation. For the warehousing system with less automation the average daily deviation between the allocated time and the time calculated by the approaches varies between 9.4 minutes and almost 3.5 hours depending on the approach. For the warehousing system with a higher degree of automation this figure lies between nine and almost 13 hours. For significant improvement the process of log design would have to be improved. The approaches have been assessed using an objective assessment scheme. From the evaluation results, possible further improvement of the approaches has been proposed, leading to recommendations for a version that could be implemented in practice.