DIY Modelling Website

These tutorial materials and videos describe the basics of do-it-yourself statistical modelling, in particular how to reproduce the R models "lm" and "glm" by coding them yourself in a maximum likelihood framework. Part 5 sets out tasks and exercises.

The materials here are written and presented by Rachel Fewster. They correspond to one week of project work for the STATS 399 BSc capstone course at the University of Auckland, New Zealand.

DIY Modelling Resources

• Presentation slides: PPTX or PDF
• Code Template (incomplete R file)
• Climate Data for task 1
• Wombat Data for task 2

Additional Resources

• Here is an interactive app by Robin Aldridge-Sutton showing maximum likelihood in action, including likelihood curves and confidence intervals for simple models.

Videos

• Part 1: DIY Modelling. What does DIY modelling mean, and what's the point of standard errors?
  

• Part 2: Numerical Optimization. How to optimize a function purely by computing it - no calculus!
  

• Part 3: DIY Bootstrap. How to write your own bootstrap code to generate standard errors.
  

• Part 4: Maximum Likelihood. What is maximum likelihood, and how do you implement it?
  

• Part 5: This Week's Task. What you need to do for this week's teamwork.
  

---

For questions or comments, please contact Rachel Fewster.