A chemometric data analysis challenge has been arranged during the first edition of the "International Workshop on Spectroscopy and Chemometrics", organized by the Vistamilk SFI Research Centre and held online in April 2021. The aim of the competition was to build a calibration model in order to predict milk quality traits exploiting the information contained in mid-infrared spectra only. Three different traits have been provided, presenting heterogeneous degrees of prediction complexity thus possibly requiring trait-specific modelling choices. In this paper the different approaches adopted by the participants are outlined and the insights obtained from the analyses are critically discussed.
翻译:在Vistamilk SFI研究中心于2021年4月在网上举办的第一版“光谱学和化学测量学国际讲习班”期间,安排了一项化学数据分析挑战,竞争的目的是建立一个校准模型,以便预测牛奶质量特征,仅利用红外光谱中所含信息;提供了三个不同的特征,显示了不同程度的预测复杂性,因此可能需要作出特定特点的建模选择;本文件概述了与会者采取的不同方法,并批判性地讨论了从分析中获得的见解。