Classification with Sparse Regularization

Here is the plan. First

Data set

Experiments

I hope that I can run and compare LR with the following regularization:

1. none
2. ridge
3. lasso
4. elastic net
5. fused lasso (total variation)
6. Random forest -- RF is the state of the art for lots of applications, but its interpretability is not very good. We also want to compare with it.

Then we can compare the accuracy, sparsity vs the robustness of each algorithm.

List of experiments