Workshop #3

Predictive modeling in R

  • Name of instructor:
    Dr. Max Kuhn(RStudio Inc.)
  • Short description:
    This workshop will contain a short introduction to predictive modeling. It will briefly cover data preprocessing, performance estimation, and modeling. A simple example is used to demonstrate these topics. Moderate experience in R and regression modeling is required for the participants.
  • Introductory background:
    Models whose main goal is to produce the most accurate prediction is the focus of this workshop. The practical aspects of modeling will be highlighted, such as visualizing data to better understand relationships between variables, preprocessing the data prior to modeling, and effectively tuning and evaluating models using resampling.
  • Tentative schedule:
    – Introductory Example
    – Exploring the Data
    – Methods for Specifying Models in R
    – Measuring Performance
    – A Simple Multinomial Model
    – Imputing Data
    – Model Tuning and Overfitting
    – A Regularized Model
    – Comparing Models