Dropout predicting is challenging analysis process which requires appropriate approaches to address the dropout. Existing approaches are applied in different areas such as education, telecommunications, retail, social networks, and banking services. The goal is to identify customers in the risk of dropout to support retention strategies. This research developed a systematic literature review to evaluate the development of existing studies to predict dropout using machine learning, following the guidelines recommended by Kitchenham and Peterson. The systematic review followed three phases planning, conducting, and reporting. The selection of the most relevant articles was based on the use of Active Systematic Review tool using artificial intelligence algorithms. The criteria identified 28 articles and several research lines where identified. Dropout is a transversal problem for several sectors of economic activity, where it can be taken countermeasures before it happens if detected early.