# Exploiting Variation for Better Decisions

## Outline.

Plan.
Concepts.
Datasets.
Questions.
Notes.
Summary.

- Provide the flavor and structure of a stat class.
- Introduce ideas and concepts.
- Introduce data.
- Use data to answer interesting questions.
- Summarize.

**Variability**.
- The simple observation that most of the data
we record is variable -- not constant.
- We need to measure/quantify the variabilty in order to exploit it.
- Quantifying uncertainty leads to an additional level of information.
- Understanding variability/precision is what differentiates statistical from
other quantitative methodologies.

**Graphical** approach to analyzing data.
- Dynamic graphics.
- Linked plots.

- Emphasize
** interpretation**, not math.

- gm92.jmp
*Daily returns on General Motors, 1992-93.* -
- gmat.jmp
*Gmat scores of the class of 94.* -

- How can we use the statistical summary of the data and a probability model
to measure
**
risk**?
- Were all MBA students subject to the same admission criteria?

- Smoothing.
- Looking for structure.
- Normal model.
- Empirical rule. 95% of data lies within +/- 2 s.d. of the mean.
- 95% of the data lies to the right of 1.645 s.d. below the mean.
- Crude VaR calculation. If mean about 0, just do 1.645 s.d. below the mean.
- For GM get VaR on $100.00 = $100 * 1.645 * 0.02 = $3.13.

- Condensed stat class.
- Interpretative, conceptual and graphical emphasis.
- Application of statistical ideas in a business context.