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
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
|
\name{dynrq}
\alias{dynrq}
\alias{print.dynrq}
\alias{print.dynrqs}
\alias{summary.dynrq}
\alias{summary.dynrqs}
\alias{print.summary.dynrq}
\alias{print.summary.dynrqs}
\alias{time.dynrq}
\alias{index.dynrq}
\alias{start.dynrq}
\alias{end.dynrq}
\title{Dynamic Linear Quantile Regression}
\description{
Interface to \code{\link{rq.fit}} and \code{\link{rq.wfit}} for fitting dynamic linear
quantile regression models. The interface is based very closely
on Achim Zeileis's dynlm package. In effect, this is mainly
``syntactic sugar'' for formula processing, but one should never underestimate
the value of good, natural sweeteners.
}
\usage{dynrq(formula, tau = 0.5, data, subset, weights, na.action, method = "br",
contrasts = NULL, start = NULL, end = NULL, ...)}
\arguments{
\item{formula}{a \code{"formula"} describing the linear model to be fit.
For details see below and \code{\link{rq}}.}
\item{tau}{the quantile(s) to be estimated, may be vector valued, but all
all values must be in (0,1).}
\item{data}{an optional \code{"data.frame"} or time series object (e.g.,
\code{"ts"} or \code{"zoo"}), containing the variables
in the model. If not found in \code{data}, the variables are taken
from \code{environment(formula)}, typically the environment from which
\code{rq} is called.}
\item{subset}{an optional vector specifying a subset of observations
to be used in the fitting process.}
\item{weights}{an optional vector of weights to be used
in the fitting process. If specified, weighted least squares is used
with weights \code{weights} (that is, minimizing \code{sum(w*e^2)});
otherwise ordinary least squares is used.}
\item{na.action}{a function which indicates what should happen
when the data contain \code{NA}s. The default is set by
the \code{na.action} setting of \code{\link{options}}, and is
\code{\link{na.fail}} if that is unset. The \dQuote{factory-fresh}
default is \code{\link{na.omit}}. Another possible value is
\code{NULL}, no action. Note, that for time series regression
special methods like \code{\link{na.contiguous}}, \code{\link[zoo]{na.locf}}
and \code{\link[zoo]{na.approx}} are available.}
\item{method}{the method to be used; for fitting, by default
\code{method = "br"} is used; \code{method = "fn"} employs
the interior point (Frisch-Newton) algorithm. The latter is advantageous
for problems with sample sizes larger than about 5,000.}
\item{contrasts}{an optional list. See the \code{contrasts.arg}
of \code{model.matrix.default}.}
\item{start}{start of the time period which should be used for fitting the model.}
\item{end}{end of the time period which should be used for fitting the model.}
\item{\dots}{additional arguments to be passed to the low level
regression fitting functions.}
}
\details{
The interface and internals of \code{dynrq} are very similar to \code{\link{rq}},
but currently \code{dynrq} offers two advantages over the direct use of
\code{rq} for time series applications of quantile regression:
extended formula processing, and preservation of time series attributes.
Both features have been shamelessly lifted from Achim Zeileis's
package dynlm.
For specifying the \code{formula} of the model to be fitted, there are several
functions available which allow for convenient specification
of dynamics (via \code{d()} and \code{L()}) or linear/cyclical patterns
(via \code{trend()}, \code{season()}, and \code{harmon()}).
These new formula functions require that their arguments are time
series objects (i.e., \code{"ts"} or \code{"zoo"}).
Dynamic models: An example would be \code{d(y) ~ L(y, 2)}, where
\code{d(x, k)} is \code{diff(x, lag = k)} and \code{L(x, k)} is
\code{lag(x, lag = -k)}, note the difference in sign. The default
for \code{k} is in both cases \code{1}. For \code{L()}, it
can also be vector-valued, e.g., \code{y ~ L(y, 1:4)}.
Trends: \code{y ~ trend(y)} specifies a linear time trend where
\code{(1:n)/freq} is used by default as the covariate, \code{n} is the
number of observations and \code{freq} is the frequency of the series
(if any, otherwise \code{freq = 1}). Alternatively, \code{trend(y, scale = FALSE)}
would employ \code{1:n} and \code{time(y)} would employ the original time index.
Seasonal/cyclical patterns: Seasonal patterns can be specified
via \code{season(x, ref = NULL)} and harmonic patterns via
\code{harmon(x, order = 1)}. \code{season(x, ref = NULL)} creates a factor
with levels for each cycle of the season. Using
the \code{ref} argument, the reference level can be changed from the default
first level to any other. \code{harmon(x, order = 1)} creates a matrix of
regressors corresponding to \code{cos(2 * o * pi * time(x))} and
\code{sin(2 * o * pi * time(x))} where \code{o} is chosen from \code{1:order}.
See below for examples.
Another aim of \code{dynrq} is to preserve
time series properties of the data. Explicit support is currently available
for \code{"ts"} and \code{"zoo"} series. Internally, the data is kept as a \code{"zoo"}
series and coerced back to \code{"ts"} if the original dependent variable was of
that class (and no internal \code{NA}s were created by the \code{na.action}).
}
\seealso{\code{\link[zoo]{zoo}},
\code{\link[zoo]{merge.zoo}}}
\examples{
###########################
## Dynamic Linear Quantile Regression Models ##
###########################
if(require(zoo)){
## multiplicative median SARIMA(1,0,0)(1,0,0)_12 model fitted to UK seatbelt data
uk <- log10(UKDriverDeaths)
dfm <- dynrq(uk ~ L(uk, 1) + L(uk, 12))
dfm
dfm3 <- dynrq(uk ~ L(uk, 1) + L(uk, 12),tau = 1:3/4)
summary(dfm3)
## explicitly set start and end
dfm1 <- dynrq(uk ~ L(uk, 1) + L(uk, 12), start = c(1975, 1), end = c(1982, 12))
## remove lag 12
dfm0 <- update(dfm1, . ~ . - L(uk, 12))
tuk1 <- anova(dfm0, dfm1)
## add seasonal term
dfm1 <- dynrq(uk ~ 1, start = c(1975, 1), end = c(1982, 12))
dfm2 <- dynrq(uk ~ season(uk), start = c(1975, 1), end = c(1982, 12))
tuk2 <- anova(dfm1, dfm2)
## regression on multiple lags in a single L() call
dfm3 <- dynrq(uk ~ L(uk, c(1, 11, 12)), start = c(1975, 1), end = c(1982, 12))
anova(dfm1, dfm3)
}
###############################
## Time Series Decomposition ##
###############################
## airline data
\dontrun{
ap <- log(AirPassengers)
fm <- dynrq(ap ~ trend(ap) + season(ap), tau = 1:4/5)
sfm <- summary(fm)
plot(sfm)
}
## Alternative time trend specifications:
## time(ap) 1949 + (0, 1, ..., 143)/12
## trend(ap) (1, 2, ..., 144)/12
## trend(ap, scale = FALSE) (1, 2, ..., 144)
###############################
## An Edgeworth (1886) Problem##
###############################
# DGP
\dontrun{
fye <- function(n, m = 20){
a <- rep(0,n)
s <- sample(0:9, m, replace = TRUE)
a[1] <- sum(s)
for(i in 2:n){
s[sample(1:20,1)] <- sample(0:9,1)
a[i] <- sum(s)
}
zoo::zoo(a)
}
x <- fye(1000)
f <- dynrq(x ~ L(x,1))
plot(x,cex = .5, col = "red")
lines(fitted(f), col = "blue")
}
}
\keyword{regression}
|
