NobleBlocks

Research index

Calibration plot.

(12969299), (12969302), (12969305) and 6 others

Figshare28 Jun 2022Open access

NobleIDark:/48914/w1/20261005/AC44F9F0DOI10.1371/journal.pone.0270548.g003

Abstract

<p>A. In the calibration plot the proportion of observed COVID-19 positives versus expected proportion of positives are plotted. Observations are grouped with an average of 50 observations per group. The expected probabilities follow from applying the inverse logit function to the CoLab-linear predictor. If the observed proportion in an external dataset is lower than the expected proportion, this means risks are over-estimated, if the observed fraction is higher, risks are under-estimated. Ideally, observed proportions are equal to expected proportions, this ideal-calibration-line is shown as a straight line through the origin with a slope of 1. The logistic calibration line is a logistic regression fit of the predicted probabilities. B. Using the intercept and/or slope from the logistic regression model, recalibrated probabilities were obtained and plotted in a second calibration plot.</p>

Discussion

Topics

More

Suggest a correction or report