Building Parametric Statistical Models
Glauco Peres da Silva, University of São Paulo
In this week long course, we will learn how to build a statistical model that captures the essential features of the process that may have generated one's observed data. Building such a model is the first step in any data analysis that will rely on parametric modelling (including for example, likelihood or Bayesian models). We will integrate a large number of common models (e.g., duration models, count models, dichotomous DV models, regression models, models of ordered data) into a single conceptual framework that will allow students to build their own new models if necessary to capture important features of the process that generated their data. We will also include a brief introduction to estimating the parameters of these models via maximum likelihood, so the week will be (somewhat) self-contained.
This course runs January 28 - February 1,2019.
Mauricio Izumi, University of São Paulo
1. Overview of building a statistical model and probability distribution
3. Basics of Calculus - Derivatives
4. Basics of probability distribution
5. Brief Introduction to Maximum Likelihood Estimation