thartbm, #rstats question (or just #statistics really): I want some goodness of fit thing for an orthogonal distance regression (or total least squares regression). Since this can apparently be done by taking the first principle component (https://stats.stackexchange.com/questions/13152/how-to-perform-orthogonal-regression-total-least-squares-via-pca) we were thinking that
Rsquared = 1 - ((sum(eigenvalues)-max(eigenvalues) / max(eigenvalues))
but we're not sure.Apparently we could also get a t-statistic or a chi-squared value: https://stackoverflow.com/questions/21395328/how-to-estimate-goodness-of-fit-using-scipy-odr
Does anyone have any opinions about this? @lakens maybe?
Add comment