Outline
Below is a rough, tentative outline of topics covered. Sometimes, topics
will be introduced simultaneously, or we will make only passing reference.
This is meant more as a study guide than a schedule. The outline gets
more vague further down the page, and our progress will depend on great part
on the interests and background of the class.
1. univariate (1 week)
a) sample distributions (descriptions)
histograms, dot-plots, boxplots, stem
and leaf
measures of center, spread
the average when used for prediction
(least squares)
b) samples as instances of populations
pdfs
means, conditional means
confidence intervals
hypothesis test review
assessing normality
2. bivariate (1 week)
a) bivariate sample
boxplots, scatterplot
mean function and variance function
lowess, smoothers
LS regression
b) population
bivariate distributions & bivariate
normal
correlation (correlation in normal
populaiton)
Regression in bivariate normal population
3. Simple Linear Regression (2 weeks)
a) The model
b) estimating parameters, slope interecept variance
c) properties of estimators
d) estimating population means and predition
f) diagnostics
i) residual plots
ii) r-squared
g) transformations
4. Multiple Regression formalities
a) Matrix review , multivariate normal review
b) The model
c) Estimating Coefficients
d) Inference
5. Multiple Regression in practice
a) Interpreting coefficients, causality and confounding
and interactions
b) assessing the model
d) relating to simple regression: added variable plots,
partial correlation
e) model selection, stepwise regression and pitfalls
6. Diagnostics
a) Leverage, influence, cross-validation, etc.
7. Principal Components Regression
8. Multi-level models