Closed mvinyard closed 1 year ago
Strongly advocate for spearman here - Pearson is redundant with MSE
On Tue, 6 Dec 2022, 12:38 pm Michael Vinyard, @.***> wrote:
What is the metric? Per suggestion by @scottgigante https://github.com/scottgigante - use a correlation (i.e., Pearson's, Spearman's) to describe the quality of low-dimension distances as they relate to the original, high-dimension differences.
How should it be implemented?
- Pearson's Correlation Coefficient https://docs.scipy.org/doc/scipy-0.14.0/reference/generated/scipy.stats.pearsonr.html
- Spearman rank-order correlation coefficient https://docs.scipy.org/doc/scipy/reference/generated/scipy.stats.spearmanr.html
Which task(s) could it be used for?
- Dimension reduction
— Reply to this email directly, view it on GitHub https://github.com/openproblems-bio/openproblems/issues/746, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACA3DX34IY5LA3KP5IQ7C2DWL52ZDANCNFSM6AAAAAASVZKLK4 . You are receiving this because you were mentioned.Message ID: @.***>
What is the metric? Per suggestion by @scottgigante - use a correlation (i.e., Pearson's, Spearman's) to describe the quality of low-dimension distances as they relate to the original, high-dimension differences.
How should it be implemented?
Which task(s) could it be used for?