Skip to content

Latest commit

 

History

History
63 lines (41 loc) · 2.5 KB

TODO.md

File metadata and controls

63 lines (41 loc) · 2.5 KB
Raw

TODO

important for first publication

  • integrate estimation of optimal lambda CvCrossManova.optimizeLambda lambda as a property of of CvCrossManova is inappropriate since it depends on the data given to runAnalyses

  • make automatic in runAnalyses, or only in ccmRegion and ccmSearchlight

  • prevent multiple reports on condition number (CvCrossManova.runAnalyses), especially in ccmSearchlight

  • Move contrast-explaining note and note on negative Ds from cvmanova.qmd to reference. Rename "API Reference" to "Notes and Reference".

  • Mention RSA in pairwise.qmd.

  • reference.qmd / helpToMarkdown: references from method / function documentation to General Notes. Implemented as a number of search strings.

  • Consistent capitalization of 'Cross-validated MANOVA' as well as 'Cross-validated Cross-MANOVA' consistent numerus for 'data'

  • integrate documentation from https://github.com/allefeld/cvmanova consider moving explanation of negative D-values from cvmanova.qmd to reference.qmd

  • check whether regressors have the same meaning across sessionsA and sessionsB, and print information labeling analyses as cross or not based on that.

  • preparation.qmd: do regions masks need to be aligned to the anatomical or the mean BOLD image?

  • Check Thomas' similarity definitions

    D = [AA AB BA BA];
test_signal_estimate(jj,1) = sqrt(abs(D(4,jj) / D(1,jj)));
pattern_similarity_estimate(jj,1) = ((D(2,jj) + D(3,jj)) / 2) / D(1,jj) /test_signal_estimate(jj,1);

    Email 2023–11–3

  • release script: zip and attach documentation HTML files

nice to have

  • give ids to errors (asserts) and warnings

  • use strings vs character arrays consistently

  • use dedicated tests instead of asserts within the tutorial see MATLAB Testing Framework https://www.mathworks.com/help/matlab/matlab-unit-test-framework.html

  • Implementation notes? classes are derived from handle & matlab.mixin.Scalar handle: prevents unnecessarily copying data and enables methods to change the object matlab.mixin.Scalar: prevents forming object arrays, for wich the classes are not designed

  • hide quasi-private methods, indicated by "_" Seems not to be necessary, except maybe checkEstimability

  • nvargs for undocumented functions: searchlight, loadDataSPM, spmReadVolMatched, spmWriteImage generally, revise undocumented functions

  • Actually, Cross-MANOVA should be called "canonical correlation".