Skip to content
No description, website, or topics provided.
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
README
common.py
convexopt_solvers.py
data_generator.py
elasticnet_eval.py
elasticnet_grid_search.py
elasticnet_hillclimb.py
elasticnet_neldermead.py
elasticnet_spearmint.py
fitted_model.py
gradient_descent_algo.py
grid_search.py
gridsearch_grouped_lasso.py
hillclimb_realdata_grouped_lasso_fullcv.py
iteration_models.py
lasso_knot_locations.py
matrix_completion_groups_eval.py
matrix_completion_groups_grid_search.py
matrix_completion_groups_hillclimb.py
matrix_completion_groups_neldermead.py
matrix_completion_groups_solver.py
matrix_completion_groups_spearmint.py
method_results.py
neldermead.py
realdata_colitis_models.py
realdata_common.py
realdata_eval.py
sgl_eval.py
sgl_grid_search.py
sgl_hillclimb.py
sgl_neldermead.py
sgl_spearmint.py
sparse_add_models_eval.py
sparse_add_models_grid_search.py
sparse_add_models_hillclimb.py
sparse_add_models_neldermead.py
sparse_add_models_spearmint.py
spearmint_algo.py

README

External packages needed:
* CVXPY is required for all scripts
* Install Spearmint https://github.com/JasperSnoek/spearmint
* Biopython is required to run the colitis data analysis. Also need to download
     * the geneset data from Molecular Signatures Database
     * GDS1615 from the Gene Expression Omnibus database

Create folders:
* In this directory, create
  * spearmint_descent directory
  * results directory
  * For each example, it will need its own results folder:
      * e.g. results/matrix_completion_groups/tmp
  * To run the Colitis data example, create a realdata directory and put the downloaded
      data into this directory

Section 3 results were generated as follows:
* Elastic Net:
    python elasticnet_eval.py
* Sparse group lasso:
    python sgl_eval.py
* Sparse Additive models:
    python sparse_add_models_eval.py

Section 4 results were generated as follows:
    python realdata_eval.py

By default, the code solves the joint optimization problem via gradient descent.
Specify different solvers (grid search, Spearmint, and nelder-mead) and problem settings via command options.
You can’t perform that action at this time.