Journal of Digital Information Management Tarik Chaghrouchni, Issam Mohammed Kabbaj, Zohra Bakkoury 16 6 2018 10.6025/jdim/2018/16/6/308-323 http://dline.info/fpaper/jdim/v16i6/jdimv16i6_3.pdf Software Process Model deviation management allows to supervise process model execution and its adaptation when an inconsisteny is detected. The process model should be adjusted based on specified rules in order to ensure the continuity of the process model execution. Some works attempted to handle process model deviation by adjusting fragments and considering some rules to ensure that all constraints are met. Other studies considered only critical constraints in order to avoid cost/ schedule overrun: rules considering the criticality of the activities and their order. However, it is still generating uncertain executions with no mastery on the final result. However, execution history can help to identify the best choice and improve the control of the model. Machine learning techniques will investigate this execution history and predict the behavior of a future execution. In this paper, we present an approach using machine learning techniques to combine the Micro-Analyze method “priority and dependency of sub-activities “ and data from previous executions to predict the appropriate software process model.