Statistics 102H: Introduction

Work load

• practice jmp by yourself (or with a friend)
• Do the examples that I don't cover in class.
• make sure you understand both HOW to do the trick I do, and WHY I do them
• Project at end of first module
• midterm at end of first module

• I won't be monitoring your work--this is an honors class, so you are on the honor system to keep up and understand.
• Feel free to find other data and look at it. If you do so, show it to me and we can discuss it.

Overview

• Three parts:
  • Prediction (via regression)
  • Science and cause and effect (via ANOVA/two-sample/regression)
  • Mathematical statistics
• Third part will be most mathematical. It will assume you have a good grasp of probabilistic models. (Say 101H, 430, or other probabiltiy background.)
• Prediction
  • Goal is forecasting, not control
  • Few assumptions--anything that works is legit
  • Criterion: how good are the fits?
  • Used in lots of science
  • New ideas coming out of AI/machine learning in computer science
• Science / cause and effect
  • Where statistics started
  • Correct attribution of blame
  • Placebo, controled experiments, etc
  • Exact same computing as prediction section--but new intepretation and more assumptions.
  • The real world creaps in here!
• Mathematical statistics
  • A bit of the math behind what we have been doing up to this point
  • Maximum likelihood estimator

First example:cleaning crews (see book)

• Follow book: this is the prediction perspective
• Science perspective: Would providing breaks improve work efficiency?
• Mathematics perspective: Show MLE computation.