Shirahama Naruki

Despite the widespread adoption of Gradient Boosting Decision Trees (GBDTs), practitioners lack systematic criteria for determining when linear models are more effective. This knowledge gap impacts model selection in applications where computational efficiency, interpretability, and extrapolation capabilities are required. This study addresses this issue through five systematic experiments that isolate data characteristics: linearity dominance, feature interactions, extrapolation requirements, small-sample scenarios, and interpretability needs. Our multi-dimensional evaluation framework integrates predictive performance with computational and interpretability costs, providing a comprehensive empirical comparison of linear regression and GBDTs. Linear models significantly outperformed GBDTs under four critical conditions.

This study aims to verify the effectiveness of a parallel learning approach for Python and R programming languages in data science education and to develop a practical learning environment using the Windows Subsystem for Linux (WSL) and Jupyter Notebook in a Bring Your Own Device (BYOD) setting. Additionally, it measures and analyzes the educational impact of this method on specialized courses in business and healthcare. The research methodology includes implementing a curriculum for simultaneous learning of both languages in the first-year "DS Programming Introduction" course, with effectiveness measured through language comprehension tests, problem-solving ability tests, and student language preference surveys. Furthermore, the study evaluates the efficacy of the BYOD environment using the WSL and Jupyter Notebook and optimizes programming education methods for specific fields. The findings of this study are expected to contribute to the development of highly skilled data science professionals by establishing new best practices in data science education.