# Sampling uncertainty of posterior probability distribution

Cross Validated Asked by Milan Bosnic on January 5, 2022

I’m working on a problem with 3 possible outcomes and a bunch of features. I have a regression model that outputs probabilities for each category and I’d like to extend these probabilities to probability distributions. So instead of getting an output of [0.2,0.3,0.5], I’d get a probability density function or at least a quantification of the uncertainty of the prediction. I’ve looked into some models that give these distributions but haven’t found any way to use it on a multiple output problem. Is there any way to do this?

Interesting problem - which is most often overlooked in data science and machine learning. The output probabilities $$bf{y}$$ are indeed estimates of the underlying (true) posterior probabilities (your $$[0.2,0.3,0.5]$$). Sampling a different training set (from your presupposed 'oracle'), will yield a slightly different set of output probabilities, when the identical input feature vector $$bf{x}$$ is presented to the classifier.

The distributions of $$hat{P}(bf{y} mid bf{x},bf{theta})$$ - they have been studied for linear and quadratic discriminant analysis ($$theta$$ is the parameter vector of the discriminant classifier).

And yes, also the sufficient parameters of these distributions of $$hat{P}(bf{y} mid bf{x},bf{theta})$$ have been derived. Specifically the variance of each posterior probability has been derived. A mathematically sound description (with the relevant references to papers in the statistical literature), can be found in Chapter 11 in the book: Discriminant analysis and statistical pattern recognition by G.J. McLachlan, Wiley (2004).

Answered by Match Maker EE on January 5, 2022

## Related Questions

### Fixed effects versus random effects in panel data for intervention group only? Change in dependent variable for each time period

0  Asked on December 11, 2021 by isobel-m

### How to calculate ARIMA(1,0,0)(1,0,1)12 prediction by hand

1  Asked on December 10, 2021 by code_diy

### Compare RMSE for the same model but varying sample size

3  Asked on December 8, 2021 by skoestlmeier

### ANOVA determining percentage of variation

2  Asked on December 8, 2021 by unistudent87

### Oscillating validation accuracy for a convolutional neural network?

3  Asked on December 8, 2021 by rockthestar

### Beta values for mixed models

1  Asked on December 8, 2021

### Comparing top level group effects using a 3-level hierarchical regression

1  Asked on December 8, 2021 by kev8484

### What are the worst (commonly adopted) ideas/principles in statistics?

32  Asked on December 8, 2021

### Statistical Analysis over different samples – Prediction for the number of objects

0  Asked on December 8, 2021

### Group level distribution for positive parameters in Bayesian multilevel models

0  Asked on December 8, 2021 by likao

### Learning more about glm parameters, how to dig deeper?

0  Asked on December 8, 2021

### OLS regression interpretation when sample means from t-test are insignificant

0  Asked on December 8, 2021 by thetagang

### Can we compare the effects of continuous covariate and categorical covariate on response variable in generalized linear regression?

0  Asked on December 8, 2021

### what is Multivariate Data

2  Asked on December 8, 2021

### Why LIME does not show the attribution for each features

0  Asked on December 8, 2021

### Is it useful or even necessary to standardize independent variables for linear regression?

1  Asked on December 8, 2021

### Why and under what conditions does Q learning converge?

0  Asked on December 8, 2021

### How to factor this conditional probability?

1  Asked on December 8, 2021 by user292136

### What are some good resources to learn Statistical Genetics?

1  Asked on December 6, 2021 by hulk