I will be grading each of the following sections
- Preprocessing
- Data cleaning
- ERP peak analysis
2 out of the following 4:
- Mass univariate
- Decoding analysis
- Source space
- Time trequency analysis
According to the following criteria:
30% Code readability, code documentation, reproducibility 30% Visualization (function over form) 30% Interpretation & results documentation 10% Sanity checks
Generally, you see that documentation is important. I do not only want to see that you analyzed the dataset, but I want to see what your motivation for choosing parameters is (like reference, filter settings, data removal criteria, algorithmic choices etc.). Regarding readability, I think direct-documentation, style and modularity are key. Regarding reproducibility, it is important that you control and document the software versions, that your analysis scripts can be easily reproduced on another computer (think about the paths you use, and if/when manual input is needed). Sanity checks are small asserts & plots here and there, maybe even a unittest or two - small things that make sure your code is doing what it is supposed to be doing.