Description Usage Arguments Details Value Note Author(s) References See Also Examples
ANOVA table calculation for objects of class "op"
1 2 | ## S3 method for class 'op'
anova(mod1, mod2, y)
|
mod1 |
null model operator matrix of class |
mod2 |
alternative model operator matrix of class |
y |
data that the models are applied to |
Calculates ANOVA as described in the reference below.
An object of class "anova"
.
This function treats the model with the larger residual sum of squares as the null model and carries out the test accordingly. With loess anova, the calculations typically are approximate since exact calculations would involve matrix products, but in this implementation, the calculations are exact since operator
already operates at this level of time complexity.
Ryan Hafen
W. S. Cleveland and S. J. Devlin. Locally weighted regression: An approach to regression analysis by local fitting. Journal of the American Statistical Association, 83(403):596–610, 1988.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 | # a simulated example with periodic component
rf <- function(x) {
n <- length(x)
sin(x * 2 * pi/200) + rnorm(n, sd = 0.5) + rep(c(0.5, 0.25,
0, -0.25, -0.5, -0.25, 0), ceiling(n/7))[1:n]
}
n <- 200
x <- c(1:n)
set.seed(8765)
ysim <- rf(x)
# null model: no seasonality
lop <- loessOp(n, span=105, degree=2)
# alternative model: seasonality
sop <- stlOp(n, n.p=7, t.window=105, t.degree=2,
s.window="periodic")
anova(sop$fit, lop, ysim)
# result:
# [1] Null model: loess with degree=2, span=105
# [2] Alt model: stl with n.p=7
# seasonal smoothing: periodic
# trend smoothing: deg=2, span=105
# l.degree=2, l.window=7, inner=2
#
# Analysis of Variance: numerator df 6, denominator df 187.09
#
# ENP RSS F-value Pr(>F)
# [1,] 6.05 68.884
# [2,] 12.05 52.751 9.5281 3.903e-09 ***
# ---
# Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
|
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.