the detailed explanations and comments of each file can be found in the README file or "READEME.pdf" in each folder.
- folder "raw_results":
all the raw results of simulations and examples, with detailed explanations;
- folder "example_test":
the simluation for the example of post-selection testing;
- folder "example_IRC":
the simulation for the IRC Example.
- folder "simul_lasso_solar":
the simulation for the lasso-solar comparison;
- folder "simul_bolasso_bsolar":
the simulation for the bootstrap selection comparison;
- folder "application":
the real-world application;
- folder "demo":
the step-by-step walkthrough of Python packages "bolasso", "solar" and "bsolar.py".
to quickly verify the paper results, all raw results are saved as HTML files in the "raw_results" folder with detailed comments and explanations.
-
we program the function for L2 error and data generating procedure based on the functions from Anaconda3 (an easy-to-use Python enviroment)
-
based on the function above, we define the function of solar, bolasso, and bootstrap solar function.
-
based on the function above, we define the functions for plotting and calculation in simulations
-
based on the function above, we conduct all simulations in ipynb files and save the raw result of simluations as HTML.
-
This package
-
was originally developed at Jan 2019 on Ubuntu 18.04 using Anaconda3 version 2019-03 (under the Intel MKL C++/Fortran library).
-
passes the unit test on Ubuntu 20.04 using Anaconda3 version 2021-04 (under the Intel MKL C++/Fortran library).
-
-
In each ".py" and ".ipynb" file I carefully and thoroughly explain
- the meaning of every step;
- the meaning of inputs and output;
- the purpose of each function;
- how each step corresponds to the paper.
-
the simulations and examples are done in ".ipynb" files;
-
all the ".py" files only contain the supporting function for simulations and examples.
-
at the end of each ".py" file, I add a testing module for debug. You can simple run each ".py" file at terminal using "python" or "python3" command. If no bug are reported, the package is bug-free.
-
The Python files automatically export the raw simulation/example results as ".html" files, which can be found at the "./raw_results" folder; the numerical results are automatically saved as ".p" files at the "numerical_result" subfolder at each simulation folder.
- To replicate the simulation after you read though the detailed explanation and comments in each ".ipynb" file, you just need to
- open each ".ipynb" file in Jupyter Lab (or Jupyter notebook),
- click the "Kernel" menu
- click "Restart Kernel and Run All Cells".
- you may want to read the comments in ".ipynb" files carefully before you replicate bolasso simulations (since it could take very long time).