SpatioTemporal (version 0.9.2)

calc.smooth.trends: Smooth Basis Functions for a mesa.data Structure

Description

A front end function for calling SVD.smooth (and SVD.smooth.cv), with either a mesa.data structure or vectors containing observations, observations times and locations. The function uses create.data.matrix to create a data matrix which is passed to SVD.smooth (and SVD.smooth.cv).

Usage

calc.smooth.trends(mesa.data = NA, obs = mesa.data$obs$obs,
    date = mesa.data$obs$date, ID = mesa.data$obs$ID, 
    subset = NA, n.basis = 2, cv = FALSE, niter = 100, 
    conv.reldiff = 0.001, df = NULL, spar = NULL)

Arguments

mesa.data

A data structure containing the observations, see mesa.data. Use either this or the obs, date, and idx inputs.

obs

A vector of observations, see mesa.data.

date

A vector of observation times, see mesa.data.

ID

A vector of observation locations, see mesa.data.

subset

A subset of locations to use when computing the smooth basis functions, see create.data.matrix.

n.basis

Number of basis functions to compute, see SVD.smooth.

cv

Also compute smooth functions using leave one out cross-validation, see SVD.smooth.cv.

niter, conv.reldiff

Controls convergence for SVD.miss.

df, spar

The desired degrees of freedom/smoothing parameter for the spline, see smooth.spline

Value

• Returns a list with the following components
• svd
  A data.frame containing the smooth trends and the dates. This can be used as the trend in mesa.data$trend.
• svd.cv
  If cv==TRUE this is list of dataframes. Each dataframe contains the smooth trend obtained when leaving one site out. Similar to SVD.smooth.cv(data)$smooth.SVD[[1]]).

  If cv==FALSE this is set to NA.

encoding

latin1

References

M. Fuentes, P. Guttorp, and P. D. Sampson. (2006) Using Transforms to Analyze Space-Time Processes in Statistical methods for spatio-temporal systems (B. Finkenst"adt, L. Held, V. Isham eds.) 77-150

See Also

Front end for calling SVD.smooth and SVD.smooth.cv, using create.data.matrix to create a data matrix from a mesa.data structure.

See also SVD.miss, svd, and smooth.spline.

Examples

##let's load some data
data(mesa.data)

##let's compute two smooth trend functions
trend <- calc.smooth.trends(mesa.data, n.basis = 2)
##and study the trends
par(mfrow=c(2,1), mar=c(2.5,2.5,.5,.5))
plot(trend$svd$date, trend$svd$V1, type="l", ylab="",xlab="",
     ylim=range(c(trend$svd$V1,trend$svd$V2)))
lines(trend$svd$date, trend$svd$V2, col=2)

##Let's exclude locations with fewer than 100 observations
IND <- names(which(table(mesa.data$obs$ID) >= 100))
##now we also compute the CV trends.
trend2 <- calc.smooth.trends(mesa.data, n.basis = 2, subset=IND,
                             cv = TRUE)

##Let's compare to the previous result
lines(trend2$svd$date, trend2$svd$V1, lty=2)
lines(trend2$svd$date, trend2$svd$V2, lty=2, col=2)

##we can also study the cross validated results to examine the
##possible variation in the estimated trends.
plot(trend$svd$date, trend2$svd$V1, type="n", ylab="",xlab="",
     ylim=range(c(trend2$svd$V1,trend2$svd$V2)))
for(i in 1:length(trend2$svd.cv)){
  lines(trend2$svd.cv[[i]]$date, trend2$svd.cv[[i]]$V1, col=1)
  lines(trend2$svd.cv[[i]]$date, trend2$svd.cv[[i]]$V2, col=2)
}

##Let's use our new trend functions in the mesa.data
mesa.data$trend <- trend2$svd