4 ms·

Seriously, though, there's one nomogram you (yes you) should know about and have it well-enough engraved in your mind's eye that you can use it with eyes closed

by cscheid 3mo ago

Seriously, though, there's one nomogram you (yes you) should know about and have it well-enough engraved in your mind's eye that you can use it with eyes closed. A nomogram for Bayes' theorem: https://www.ovid.com/journals/nejm/abstract/10.1056/nejm197507312930513~nomogram-for-bayess-theorem?redirectionsource=fulltextview https://www.ovid.com/journals/nejm/abstract/10.1056/nejm1975...

senkora 3mo ago
That is cool, although it took me awhile to understand it because the posterior probability is on the left and the prior probability is on the right, and because it uses D=Disease and T=Test when I am used to seeing D=Data.
kqr 3mo ago
Neat. This is based on Bayes' rule in its odds form[1], or more specifically in log-odds form, where evidence is additive[2]. [1]: https://entropicthoughts.com/bayes-rule-odds-form https://entropicthoughts.com/bayes-rule-odds-form [2]: https://entropicthoughts.com/sensitivity-counts-against-you https://entropicthoughts.com/sensitivity-counts-against-you
- riedel 3mo ago
  Actually I find nomograms in log form really cool for making naive bayes classifiers 'explainable'. One can even add density for continuous values. IMHO this is so much nicer than e.g. decisions tree visualizations (which everyone quotes for the most explainable AI models).
  - tgv 3mo ago
    It is indeed a great tool for visualizing Bayesian relations. You can even "feel" the sensitivity.
speff 3mo ago
That was a bit small on my screen. Found an interactive one here that's scalable - https://www.medcalc.org/en/calc/fagans-nomogram.php https://www.medcalc.org/en/calc/fagans-nomogram.php
trunch 3mo ago
Can you use actually use it eyes closed? Never heard of that level of precision in the mind's eye
- cscheid 3mo ago
  You're right that I can't reproduce it like a phone camera could, and that the more precise you are the better (and it might be that with my eyes closed I'm doing something closer to addition in log-odds, like a separate comment responded), but this is super useful even coarsely. The visual affordance gives Bayes's theorem a physicality unlike any other tool I've found.

by cscheid 3mo ago

senkora 3mo ago

That is cool, although it took me awhile to understand it because the posterior probability is on the left and the prior probability is on the right, and because it uses D=Disease and T=Test when I am used to seeing D=Data.

kqr 3mo ago

Neat. This is based on Bayes' rule in its odds form[1], or more specifically in log-odds form, where evidence is additive[2]. [1]: https://entropicthoughts.com/bayes-rule-odds-form https://entropicthoughts.com/bayes-rule-odds-form [2]: https://entropicthoughts.com/sensitivity-counts-against-you https://entropicthoughts.com/sensitivity-counts-against-you

riedel 3mo ago
Actually I find nomograms in log form really cool for making naive bayes classifiers 'explainable'. One can even add density for continuous values. IMHO this is so much nicer than e.g. decisions tree visualizations (which everyone quotes for the most explainable AI models).
- tgv 3mo ago
  It is indeed a great tool for visualizing Bayesian relations. You can even "feel" the sensitivity.

speff 3mo ago

That was a bit small on my screen. Found an interactive one here that's scalable - https://www.medcalc.org/en/calc/fagans-nomogram.php https://www.medcalc.org/en/calc/fagans-nomogram.php

trunch 3mo ago

Can you use actually use it eyes closed? Never heard of that level of precision in the mind's eye

cscheid 3mo ago
You're right that I can't reproduce it like a phone camera could, and that the more precise you are the better (and it might be that with my eyes closed I'm doing something closer to addition in log-odds, like a separate comment responded), but this is super useful even coarsely. The visual affordance gives Bayes's theorem a physicality unlike any other tool I've found.