"The number of variances > The number of samples"
, it is called "Sparse".
Sparce modeling is the one of the method of Selection of Variance. It select variances automatically.
The idea of sparce modeing is that "Final mode is made by small number of variances."
Lasso regression is the method that it uses the idea of sparce modeling for Multi-Regression Analysis.
Principal Component Regression Analysis uses the different way to improve the weakness of multi-regression analysis.
Graphical Lasso is the method that it uses the idea for Correlation Analysis for Multi-Variable.