ScholarWorks@숙명여자대학교

초록

Although existing machine learning solutions make it easy to analyze data, extracting the interesting parts from massive files and merging them is still a manual process, and the greater the number of distributed files, the more burden this places on researchers. This study proposes a web-based machine learning platform that automatically extracts data from files using regular expressions and analyzes them using machine learning. The proposed system provides a user-friendly GUI allowing users to easily train model and evaluate and visualize the performance. To validate the effectiveness it was applied to the study of predicting the properties of lithium-ion battery materials and it shows that R² and Root Mean Squared Logarithmic Error(RMSLE) were 0.954 and 0.094, respectively. This result reveals that researchers without knowledge about data engineering can adopt machine learning using the proposed system.

키워드