ML software systems, like those used in machine learning, often undergo repetitive changes. While tools exist for Java, there is a lack of such tools for Python, which is widely used in ML development. To address this gap, Java static analysis tools were adapted for Python, allowing for a detailed study of code change patterns in 59 ML systems. The adapted tools, RefactoringMiner and CPATMiner, revealed actionable insights for tool builders and were released for further research.