In various industrial sectors, data-driven machine system condition diagnosis is crucial for maintaining machine performance and safety, making it essential to analyze data patterns and accurately understand the state of machines. Scatter plots, which are commonly utilized as visualization techniques, are employed for data pattern analysis. However, in complex systems with numerous features, creating and analyzing scatter plots for all feature pairs becomes impractical. Therefore, this study proposes a method for automating scatter plot creation and feature selection based on Euclidean distance. This approach efficiently identifies critical features in the data analysis process, ensuring consistency and accuracy in variable selection and is expected to contribute to machine system condition diagnosis and performance optimization.
