We are excited to invite you to our bi-monthly Meetup. This session will focus on the basics of Bayesian statistics.
• Doors open at 18:00; food will be provided
• Talk starts at 19:00
• Drinks as of 20:00
Bayesian statistics is tailored towards understanding and predicting properties such as click rates on banners, the likelihood of costumers making certain decisions or fraud being committed.
Furthermore, Bayesian methods of statistical inference circumvent certain problems in traditional (or frequentist) statistics, such as the difficulties with significance testing when only little data is available. Through a series of examples, the approach of traditional statistics is contrasted with Bayesian statistics, while outlining pros and cons of the respective approaches.
As an introduction, we’ll start by rehashing basics of Bayes’ law from some pedestrian, but nonetheless important, examples. For instance, how can Bayes’ law help in the context of judicial errors and sentiment analysis of texts? However, for many application purposes, such as source separation in the cocktail party problem, one needs advanced numerical methods in order to obtain predictions from Bayes’ law. We’ll address methods such as a stochastic gradient algorithm and Monte Carlo simulations.
Prerequisite: Knowledge of basic statistics