-
Data
the basic score os observations which we want to analyze
-
Model
- makes a specific prediction for each observation in the data, which can vary in complexity
- basic models are tailored to fit given circumstances
- ideal model is parsimonious with many fewer parameters than observations
-
Error
- the amount by which the model fails to represent the data accurately
- a measure of how much the model mispredicts what is actually observed
- can be decreased by adding parameters to the model so that predictions are conditional of additional information to each observation
-
Parameter
a numerical characteristic of a population which is hypothesized to be a significant predictor of the data
-
Model C
- Compact Model
- null hypothesis
-
Model A
- Augmented Model
- alternative hypothesis
-
Proportional Reduction in Error
- PRE
- index, as a percentage, of how worthwhile additional parameters are to the model
- values are between 0 and 1, where values closer to 1 are more worthwhile
-
-
Ýi
prediction of ith observation
-
Xij
observed value of j for ith parameter
-
-
εi
true amount of difference between Yi and β0, if β0 is known exactly
-
bi
estimated model parameter
-
ei
- amount that predictions misses the actual observation
- an estimate of εi
- measures of variability
- measures of spread
-
βiXij
adjustment of basic prediction
-
β0
unknown parameter estimated from the data
-
b0
- estimate of β0 that is derived from the data
- measure of location
- measure of central tendency
|
|