The advent and ongoing exponential growth of data-driven systems have led to the adoption of machine learning (ML) and artificial intelligence (AI) related systems by organizations. While it is inviting organizations to build machine learning systems to leverage data and reap prediction benefits out of it, there are fallacies and false solutions that guide organizations and practitioners into applying poor practices in the development of such machine learning systems. Implementing such bad practices degrades the system, leads to technical debt and makes it harder to maintain and get the best out of it. Using the concepts of machine learning systems and antipatterns, the thesis first presents the definition of machine learning system antipattern. Subsequently, using a catalog of antipatterns in machine learning systems, the thesis provides a way to inform practitioners and machine learning engineers about these bad practices and false patterns, so they can counteract them. For better visualization and documentation, the catalog itself is exported to a web application. The catalog presented is extensible and can further be extended by practitioners with more antipatterns covering new aspects of machine learning systems.