A L 1-norm PCA and a Heuristic Approach
A L 1-norm PCA and a Heuristic Approach
复制标题
DOI:
10.1007/978-3-642-61159-9_32
复制
发表时间:
1996
期刊:
影响因子:
--
通讯作者:
A. Baccini;Philippe C. Besse;A. Falguerolles
中科院分区:
文献类型:
--
作者:
A. Baccini;Philippe C. Besse;A. Falguerolles
A model for a L1 principal component analysis (PCA) is considered and discussed. On the other hand, a PCA based on Gini's mean absolute di erences is introduced in order to obtain heuristic estimates of the L1 model. This PCA is connected to the canonical correlation analysis of the original data and their ranks. The method is illustrated on real and arti cial data with the aim of achieving proper data reduction.