# Calibration plots

**URL:** <https://discourse.datamethods.org/t/calibration-plots/5912>\
**Category:** models\
**Tags:** calibration\
**Created:** [August 27, 2022, 9:24am UTC](https://discourse.datamethods.org/t/calibration-plots/5912 "2022-08-27T09:24:42Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![KirovDoc](https://discourse.datamethods.org/user_avatar/discourse.datamethods.org/kirovdoc/32/3178_2.png) [@KirovDoc](https://discourse.datamethods.org/u/KirovDoc)\
**Post date:** [August 27, 2022, 9:24am UTC](https://discourse.datamethods.org/t/calibration-plots/5912/1 "2022-08-27T09:24:42Z")

</div>

Hi! I need to ask about calibration again. I created the LR model and did a calibration plot on the test data (n=49). Below are two variants:  
Here is the first calibration plot using Caret. Looks like a true story!  
 ![1](https://discourse.datamethods.org/uploads/default/original/2X/b/bf02f59fc531d67481066d14582d37757d9ea94c.png)  
The second is calibration plot using Rms, val.prob (predicted probability, class). The actual probability is odd, and it looks different. Is there something I’m doing wrong?  
 ![2](https://discourse.datamethods.org/uploads/default/original/2X/3/315ae6d33d4ceb128706057d599dfb004a685fd3.png)

---

<div class="post-metadata">

**Author:** ![f2harrell](https://discourse.datamethods.org/user_avatar/discourse.datamethods.org/f2harrell/32/165_2.png) [@f2harrell](https://discourse.datamethods.org/u/f2harrell)\
**Post date:** [August 27, 2022, 11:23am UTC](https://discourse.datamethods.org/t/calibration-plots/5912/2 "2022-08-27T11:23:35Z")

</div>

Please post this as a reply at the end of [RMS Describing, Resampling, Validating, and Simplifying the Model](https://discourse.datamethods.org/t/rms-describing-resampling-validating-and-simplifying-the-model/4790) and I’ll remove it here. Note that binning is not valid in this context so re-do your own calibration curve without binning before posting. If you are unable to do this, re-post it as-is.