Latin-partition主要是一种交叉验证方法，由professor Peter及其课题组开发，用于模型的稳定性和有效性验证评价,，可以作为自助多次重复划分校正集和验证集，避免重复随机数据集建模所产生的不可靠性。-Latin-partition is primarily a cross-validation method, professor Peter and developed by the Task Force for the stability and validity evaluation model can be classified as self-calibration set and repeated validation set, to avoid duplication of random data the resulting set of modeling unreliability.