This sub-directory include the code and notebooks used to analyze the results from the main experiment as well as make plots and tables, etc... Included below is a brief description of what each included file is.
-
Notebook exploring how results change when limited to only different clusters of target variables.
-
Notebook dedicated to the creation of extra supplementary figures addressing some sub-questions of interest.
-
File including helper utilities for loading in raw saved numpy results into an organized form.
-
Notebook where the various interactive plots (created with plotly) are generated and saved as html files.
-
Notebook with an introduction on how to load and work with the raw results, for those interested in performing their own analyses on the raw results from this study.
-
Notebook responsible for generating the final, neat / pretty figures which will appear in the main manuscript for this work.
-
Notebook responsible for generating the HTML versions of different results tables featured throughout the online project documentation.
-
Notebook dedicated to answering the question on how a front-end feature selection scheme may or may not influence parcellation-performance scaling. Hint: it doesn't.
-
Notebook dedicated to exploring how runtime differed across different combinations of ML pipeline, ensemble strategy and parcellation.
-
The main file containing plotting functions used to generate and summarize results, used across most notebooks within this directory.
-
Notebook dedicated to exploring a subset of special ensembles created from collections of existing parcellations.
-
Notebook containing the analyses code for all main statistical analyses included in the study, using library statsmodel.
-
Notebook where all of the target variables are automatically used to generated a summary .docx table.