# Author Archives: abenavoli

## The importance of the region of practical equivalence (ROPE)

The difference between two classifiers (algorithms) can be very small; however there are no two classifiers whose accuracies are perfectly equivalent. By using an null hypothesis significance test (NHST), the null hypothesis is that the classifiers are equal. However, the null hypothesis is practically always false! By rejecting the null …

## Bayesian Signed-Rank Test for comparing algorithms in machine learning

This post is about Bayesian nonparametric tests for comparing algorithms in ML. This time we will discuss about Python module signrank in bayesiantests (see our GitHub repository). It computes the Bayesian equivalent of the Wilcoxon signed-rank test. It return probabilities that, based on the measured performance, one model is better …

## 28 October, USA general election situation

These are the updated probability obtained running the Python code that computes the Bayesian posterior distribution over the electoral votes using near-ignorance priors. The worst and best case distribution for Clinton are in red and respectively, blue.. Winning probability above 0.99 (for both worst and best scenario). Electoral votes between …

## Quantum Fréchet bounds

If you are interested in the Quantum Mechanics version of Fréchet bounds, then I have just edited the Fréchet inequalities page in wikipedia to show that similar bounds can also be obtained in quantum mechanics for separable density matrices. These bounds were derived in our paper: It is worth to …

## Bayesian Sign Test

Bayesian Sign Test Module signtest in bayesiantests computes the probabilities that, based on the measured performance, one model is better than another or vice versa or they are within the region of practical equivalence. This notebook demonstrates the use of the module. We will load the classification accuracies of the …

## Fresh forecast for US2016 election

## US2016 election forecast

I have run again the Bayesian algorithm that uses a prior near-ignorance model to compute US2016 election forecast. This is the current situation for Clinton (worst-case in red and best-case in blue). The probability range of winning the election (by getting the majority of the electoral votes) is [0.68,0.91]. The …

## ECML 2016 tutorial on Bayesian vs. Frequentist tests for comparing algorithms

## Clinton vs. Trump 23th Sptember 2016

I have run again the Python code that computes the worst-case (red) and best-Case (blue) posterior distribution for Clinton winning the general USA election. using fresh (September) poll-data. At the moment there is a quite large uncertainty but is still in favour of Clinton: the probability of winning …