Designing a software model using Unified Modeling Language (UML) does not fully integrate some essential model elements, including goals and non-functional requirements. However, it can extend model elements and relationships through the UML profile to develop model components. In this study, we propose a goal-centralized metamodel that explicitly captures functional and non-functional requirements-based goal-oriented analysis for machine learning systems in an integrated manner. This study aims to present the integration approaches to specific ML model elements that focus on modeling and analyzing goals and scenarios.