Seminário

What On Earth Is Bayesian Statistics And Why Do I Keep Hearing About It?

Sala 6.4.30, Ciências ULisboa (com transmissão via Zoom)

Por Rosina Savisaar (Mondego Science).

For many of us, the desperate struggle for significance is what analyzing data boils down to in practice. But is it possible to do statistics differently? May it perhaps even be better to do statistics differently? P-values, confidence intervals and statistical tests are primarily associated to a school of statistics referred to as frequentism. In the vast majority of introductory courses, this is what is taught. Many of us never get exposed to anything else. Although this paradigm is not without its merits, there is also an alternative school of thought, known as Bayesian statistics. Bayesian statistics is a way of analyzing data that was routine during the 19th century, fell into disgrace for most of the 20th, and has seen a remarkable revival during the 21st. For decades, proponents of this paradigm have claimed that it is more intellectually coherent, more powerful and easier to learn for beginners than its frequentist alternative. In this talk, we will walk through a simple example of how data analysis works in this paradigm, in a way that should be accessible also to complete beginners in statistics. In the process, we will try to understand how this way of looking at data and probabilities differs from the frequentist approach. Finally, I will argue that there is much to gain in introducing these methods from the very start, in the very first introductory statistics courses that researchers take during their training.


Transmissão via Zoom.

14h30
CEAUL - Centro de Estatística e Aplicações da Universidade de Lisboa / CEMAT - Centro de Matemática Computacional e Estocástica
Título e data do evento, inseridos em fotografia de cinco jovens em contexto de investigação

Pré-inscrições já disponíveis!

Título do prémio

As candidaturas decorrem até ao dia 31 de maio.

Título/data do evento, logótipos das entidades organizadoras e fotografia de Lisboa (Castelo de S. Jorge e respetiva colina)

Inscrição (taxa reduzida) até 20 de março.

Título/data/local do evento, logótipos das entidades organizadoras e várias fotografias da orla costeira e de pessoas

Inscrições preferencialmente até 01 de março.

Logótipo do evento, sobre um fundo branco

Um evento de reunião da comunidade nacional nas diversas vertentes da informática, com a ambição de ser o fórum de eleição para a divulgação, discussão e reconhecimento de trabalhos científicos.

Páginas