Chapter 7 (Time series regression models) Flashcards
1
Q
Linear regression
A
Linear relationship between forecast variable ‘y’ and predictor variables ‘x’
2
Q
Linear regression assumptions about errors
A
- mean zero
- not autocorrelated
- unrelated to predictor variables
- (useful) to have errors normal distributed
3
Q
Least square estimation
A
finds the best estimates of coefficients
4
Q
Coefficient of determination / R-squared
A
how well a linear regression model fits the data
Lies between 0 to 1
5
Q
Evaluating the regression model
A
- ACF plot of residual will likely show autocorrelation, still unbiased but will have larger prediction intervals (and short term forecasts may not be accurate)
- Histogram of residuals are normally distributed
- Residual plots against predictors should not show patterns
- Residual plots against fitted values should show no pattern
- Outliers
6
Q
Use of Fourier terms
A
Regression model containing Fourier terms is often called harmonic regression
7
Q
Piecewise linear
A
Introduce knots (points) where slope of f can change