Minimal dataset and code to run the original MATLAB implementation of the ridge regression encoding model.
This code is tested on MATLAB R2021b (and is known to NOT work on R2022a).
Access the example dataset here and unzip it.
I have also uploaded some saved models (the output of ridgeModel_saveResults.m) that can be used for direct benchmarking. Those can be found here.
Edit line 5 of main.m with the with the path to the downloaded data, then run main.m for an example of how the encoding model is fit. This is the only line that should need modification for the code to run.
There is a lot of refactoring we can do here, especially in ridgeModel_returnDesignMatrix.m, but I figure we can save that work for the Python implementation.
Todo:
- add code to generate plots of cvR^2^ and beta kernels
- tests to compare design matrices and model predictions
- coefficient of determination calculation
Feel free to slack Max Melin if anything is unclear or not running as expected.