MP2-Method: A Visual Analysis Method for Material Performance Prediction Supporting Model Selection
Qikai Chu, Yadong Wu, Lingli Zhang, Pan Li, Zhenjin Song, Hui Pu, Zhouqiao He, Weihan Zhang · 2022 5th International Conference on Pattern Recognition and Artificial Intelligence (PRAI) · 2022
With the advent of Industry 4.0, the traditional way of fluorine material research by adjusting parameters through multiple trials can no longer meet the demand of multi-source heterogeneous and large amount of complex data processing and analysis, the visual analysis technology based on machine learning models brings opportunities for material science research. However, due to the high confidentiality of fluorine material data, it is not convenient for factories to trust the data to third-party professionals for processing and analysis. Therefore, this paper introduces a visual analysis method for material performance prediction supporting model selection, MP2-Method, for different data sets, which supports researchers’ autonomous selection and comparison of different levels of prediction models, and enables performance prediction of fluorine materials using visual analysis by adjusting control parameters. In addition, the method introduces uncertainty visual analysis to reduce the uncertainty of the control parameter data and improve the prediction accuracy. Finally, the usefulness and reliability of the MP2-Method is demonstrated by case studies and interviews with domain experts.