Reconfiguration of machine tooling components in industrial production systems is highly dependent on the expertise of human operators. This expertise is developed over time through the experience of skilled workers. A significant challenge is the absence of a predefined set of actions and their sequences. Moreover, due to variability in user behavior, structured tasks are often performed with diverse patterns and sequences that differ across users and contexts. This work analyzes the variability in reconfiguration patterns caused by differences in action sequences of an operator, by employing an action prediction model that annotates action and sequence semantically. The findings are evaluated based on experts’ action duration and sequence. Results highlight the potential to identify skillsets for reconfiguring machine tooling in forming processes use-case, paving a way for systems to enhance their productivity on the shopfloor.