Write markdown files which will then be "compiled" into ipynb using pandoc. (see e.g. this example.)
The notebook reproducing the original paper:
Introduction: Introduce the task and the original result we are trying to reproduce
Import data: (Corresponds to "feature extraction" section in paper.) Import the already-processed data from CSV files, but also explain how to re-create the features if need be. Do a little bit of exploratory data analysis, e.g. count the number of samples in each class and compare to what the original paper says (262 and 38, respectively.)
Oversampling: (Corresponds to "data balancing" section.) You can explain a little bit about what ADASYN does, make sure to cite sources.
Train model: only the one with the best results, according to the original paper. Use 10-fold CV for evaluation. Since the original paper does not report hyperparameters, we have to decide what to do about that.
Evaluate: report accuracy, sensitivity, specificity and error, and confusion matrix for the best model.
Write markdown files which will then be "compiled" into ipynb using
pandoc. (see e.g. this example.)The notebook reproducing the original paper:
Then, repeat with correct oversampling.