I'm working on writing an excel program that logs a series of measurements. The measurements are then used in 6 different models, all approximating the same unknown value.

I'd like to take the result of these 6 models and approximate the most probable value for what the unknown is. I've used average and medical average but given the range in values of my set, I don't think these are necessarily appropriate.

I've checked out several other similar questions on this stack, but the solutions seem to be written for higher level programming languages than I have access to. I'm not sure what the math is that I'm looking for, but I remember something similar to this from my undergrad courses. And I seem to remember the math being relatively easy to compute by hand. Ideally I'd like to be able to translate this into an excel style solution.



closed as unclear what you're asking by Glen_b Apr 16 '17 at 14:49

Please clarify your specific problem or add additional details to highlight exactly what you need. As it's currently written, it’s hard to tell exactly what you're asking. See the How to Ask page for help clarifying this question. If this question can be reworded to fit the rules in the help center, please edit the question.

  • $\begingroup$ Thanks for the down vote without a word of what you don't understand. Very helpful $\endgroup$ – Diesel Apr 16 '17 at 18:25
  • $\begingroup$ unfortunately it (this sort of downvoting) is happening a lot just now; someone is indiscriminately downvoting posts -- especially from newer users -- without commenting. Frequently there's nothing discernably wrong.. In your case however, your post is very unclear (which is why I put it on hold, and you can't have failed to notice that) and needs to be made clearer (but the downvote was already there when I saw it). I'd have upvoted your post if it wasn't quite so unclear. Among things you'll need to clarify: what's a "medical average"? Why take any kind of averages at all? ... ctd $\endgroup$ – Glen_b Apr 17 '17 at 1:42
  • $\begingroup$ ctd... How are you defining "most probable" here -- what is your probability model over the approaches to estimation (/models for the measurements)? Is this intended to be some Bayesian model selection framework and if not, how is "most probable" intended? Do you instead mean to refer to some comparison of likelihoods over models? How, in short, are we to guess what you want? I have a fair bit of experience with unclear questions but I really can't even properly guess what's going on in your post. $\endgroup$ – Glen_b Apr 17 '17 at 1:42
  • $\begingroup$ I have concerns that maybe the answer doesn't really answer what you might be asking, but even if you feel that it does, in its present state it can't be much help to others. A few clarifications would likely make it much more useful to later readers. $\endgroup$ – Glen_b Apr 17 '17 at 1:47

If you have several estimators of the same value, you can combine them by using a weighted average of all the estimates where the weights are given based on the variance of each estimator. Check for example: https://en.m.wikipedia.org/wiki/Inverse-variance_weighting

  • $\begingroup$ This looks like exactly what I'm looking for, thank you! $\endgroup$ – Diesel Apr 16 '17 at 18:32

Not the answer you're looking for? Browse other questions tagged or ask your own question.