Skip to content

Corrected model-to-ceiling mapping under explained variance #2169

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
sandervanbree wants to merge 1 commit into
base: master
Choose a base branch
from
Open

Corrected model-to-ceiling mapping under explained variance #2169

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
9 changes: 3 additions & 6 deletions brainscore_vision/benchmark_helpers/neural_common.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -40,12 +40,9 @@ def timebins_from_assembly(assembly):


def explained_variance(score: Score, ceiling: Score) -> Score:
# ro(X, Y)
# = (r(X, Y) / sqrt(r(X, X) * r(Y, Y)))^2
# = (r(X, Y) / sqrt(r(Y, Y) * r(Y, Y)))^2 # assuming that r(Y, Y) ~ r(X, X) following Yamins 2014
# = (r(X, Y) / r(Y, Y))^2
r_square = np.power(score.values /
ceiling.values, 2)
# Note: ceiling here is the split-half reliability r(Y, Y), which bounds model explained variance.
# Thus, model-to-ceiling normalization first requires bringing score (r) into r^2 before dividing it by the ceiling (see https://osf.io/preprints/psyarxiv/gjk45).
r_square = np.power(score.values, 2) / ceiling.values
ceiled_score = Score(r_square)
if 'error' in score.attrs:
ceiled_score.attrs['error'] = score.attrs['error']
Expand Down