Rule-Based Chatbot System Design for Decision-Making Processes: A Structured Approach to Statistical Analysis Selection

In this study, a rule-based chatbot system based on a Finite State Machine (FSM) was designed, devel-oped, and piloted to support the selection process for multivariate statistical analysis. Uncertainties and methodological errors encountered in statistical analysis selection make the decision-making process difficult for researchers and necessitate systematic support for this process. The system developed in this context follows a deterministic decision flow based on structured user inputs. It provides guidance on appropriate statistical analysis methods, taking into account factors such as the dependent variable type, data structure, and research objective. The system architecture consists of a user interface, dialogue management, an FSM-based decision engine, a statistical knowledge base, and output generation components. The system was developed following a waterfall model, and the analysis selection process was structured using a decision tree with finite states and transition logic. Within the scope of the pilot implementation, assumption checks for linear regression analysis were modeled using rule-based methods, and a web-based interactive system prototype was developed. The developed system was evaluated by five experts in statistics and data analytics on usability, clarity of decision logic, and perceived benefit. Expert evaluations show that the system has a practical, structurally consistent, and scalable architecture. These findings indicate that rule-based chatbot approaches are a suitable solution for decision-support scenarios where explainability and consistency are paramount.