解释这个r代码在干什么: ----------------- Climate zone effects for temperature -----------------…
- User
解释这个r代码在干什么:
----------------- Climate zone effects for temperature -----------------
Section 10.1.1 Climate Region Effects on Temperature
library(fda)
set up the data for the analysis
CanadianWeatherregion==regions.[1],1] regions. = unique(CanadianWeather$region) regions.
find how many cities in each region
Atlantic = CanadianWeatherregion==regions.[1],1] dim(Atlantic) Continental = CanadianWeatherregion==regions.[2],1] Pacific = CanadianWeatherregion==regions.[3],1] Arctic = CanadianWeatherregion==regions.[4],1]
quartz() par(mfrow=c(1,1),mar = c(8, 8, 4, 2)) matplot(Atlantic,xlab="Day", ylab="Mean Temperature", col="red",type="l",ylim=c(-40,25),cex.lab=2,cex.axis=2) for (j in 1:ncol(Continental)) {lines(Continental[,j],xlab="Day", ylab="Mean Temperature",col="blue",type="l")} for (j in 1:ncol(Pacific)) {lines(Pacific[,j],xlab="Day", ylab="Mean Temperature",col="black",type="l")} for (j in 1:ncol(Arctic)) {lines(Arctic[,j],xlab="Day", ylab="Mean Temperature",col="brown",type="l")} legend('topright', regions., lty=1, col=c('red', 'blue', 'black','brown'), bty='n', cex=.75)
p = length(regions.) + 1 regionList = vector("list", p) names(regionList)= c('Canada', regions.) regionList[[1]] = c(rep(1,35),0) for (j in 2:p) { xj = (CanadianWeather$region == regions.[j-1]) regionList[[j]]= c(xj,1) }
Smoothing on the temperature data
Lcoef = c(0,(2*pi/365)^2,0) harmaccelLfd = vec2Lfd(Lcoef, c(0,365))
tempbasis = create.fourier.basis(c(0, 365), 65)
lambda = 1e6 tempfdPar65 = fdPar(tempbasis, harmaccelLfd, lambda)
temp = dailyfd quartz() par(mfrow=c(1,1),mar = c(8, 8, 4, 2)) plot(tempfd,xlab="Day", ylab="Mean Temperature",lwd=2,cex.lab=2,cex.axis=2)
augment tempfd by adding a 36th observation with temp(t) = 0
coef = tempfdfdnames)
set up the regression coefficient list
betabasis = create.fourier.basis(c(0, 365), 11) betafdPar = fdPar(betabasis) betaList = vector("list",p) names(betaList)= regions. for (j in 1:p) betaList[[j]] = betafdPar
carry out the functional analysis of variance
fRegressList= fRegress(temp36fd, regionList, betaList)
extract the estimated regression coefficients and y-values
betaestList = fRegressListyhatfd regions = c("Canada", regions.)
Figure 10.1
quartz() par(mfrow=c(2,3)) plot(betaestList[[1]]fd, lwd=2, xlab="Day", ylab="Difference", main=regions[j],cex.lab=2,cex.axis=2)
--------------- Testing for no effect of climate zone ----------------
temp36fd, regionList, betaList from Section 10.1.1 above
F.res = Fperm.fd(temp36fd, regionList, betaList) names(F.res)
Figure 10.14
plot in black and white
quartz() par(mfrow=c(1,1),mar = c(8, 8, 4, 2))
with(F.res,{ q = 0.95 ylims = c(min(c(Fvals, qval, qvals.pts)), max(c(Fobs, qval))) plot(argvals, Fvals, type = "l", ylim = ylims, col = 1, lwd = 2, xlab = "day", ylab = "F-statistic", cex.lab=1.5,cex.axis=1.5) lines(argvals, qvals.pts, lty = 3, col = 1, lwd = 2) abline(h = qval, lty = 2, col = 1, lwd = 2) legendstr = c("Observed Statistic", paste("pointwise", 1 - q, "critical value"), paste("maximum", 1 - q, "critical value")) legend(argvals[1], 1.4, legend = legendstr, col = c(1, 1, 1), lty = c(1, 3, 2), lwd = c(2, 2, 2)) } )
- Assistant
这段R代码主要是用来分析加拿大的不同气候区对温度的影响,具体步骤如下:
1. 设置必要的环境
library(fda)加载功能数据分析库
fda。2. 准备和初始化数据
CanadianWeather$dailyAv[,CanadianWeather$region==regions.[1],1] regions. = unique(CanadianWeather$region) regions.在加拿大天气数据
CanadianWeather里,提取每个气候区的日平均温度,并找到不同气候区的标签regions.。3. 提取每个气候区的城市数
分别提取
Atlantic,Continental,Pacific, 和Arctic四个气候区的气象数据,并检查数据维度:Atlantic = CanadianWeather$dailyAv[,CanadianWeather$region==regions.[1],1] dim(Atlantic) Continental = CanadianWeather$dailyAv[,CanadianWeather$region==regions.[2],1] Pacific = CanadianWeather$dailyAv[,CanadianWeather$region==regions.[3],1] Arctic = CanadianWeather$dailyAv[,CanadianWeather$region==regions.[4],1]4. 可视化每个气候区的温度曲线
quartz() par(mfrow=c(1,1),mar = c(8, 8, 4, 2)) matplot(Atlantic, xlab="Day", ylab="Mean Temperature", col="red", type="l", ylim=c(-40,25), cex.lab=2, cex.axis=2) for (j in 1:ncol(Continental)) { lines(Continental[,j], xlab="Day", ylab="Mean Temperature", col="blue", type="l") } for (j in 1:ncol(Pacific)) { lines(Pacific[,j], xlab="Day", ylab="Mean Temperature", col="black", type="l") } for (j in 1:ncol(Arctic)) { lines(Arctic[,j], xlab="Day", ylab="Mean Temperature", col="brown", type="l") } legend('topright', regions., lty=1, col=c('red', 'blue', 'black', 'brown'), bty='n', cex=.75)将每个气候区的温度曲线绘制在一张图上,并使用不同颜色进行区分。
5. 准备功能数据分析
Lcoef = c(0,(2*pi/365)^2,0) harmaccelLfd = vec2Lfd(Lcoef, c(0,365)) tempbasis = create.fourier.basis(c(0, 365), 65) lambda = 1e6 tempfdPar65 = fdPar(tempbasis, harmaccelLfd, lambda) temp = daily$tempav day = 1:365 tempSmooth65 = smooth.basis(day, temp, tempfdPar65) tempfd = tempSmooth65$fd quartz() par(mfrow=c(1,1),mar = c(8, 8, 4, 2)) plot(tempfd, xlab="Day", ylab="Mean Temperature", lwd=2, cex.lab=2, cex.axis=2)对温度数据进行平滑处理,生成功能数据对象
tempfd,并对其进行可视化。6. 构建回归模型
coef = tempfd$coef coef36 = cbind(coef, matrix(0, 65, 1)) temp36fd = fd(coef36, tempbasis, tempfd$fdnames) betabasis = create.fourier.basis(c(0, 365), 11) betafdPar = fdPar(betabasis) betaList = vector("list", p) names(betaList) = regions. for (j in 1:p) betaList[[j]] = betafdPar fRegressList = fRegress(temp36fd, regionList, betaList) betaestList = fRegressList$betaestlist regionFit = fRegressList$yhatfd将温度数据扩展并构建基于傅里叶基函数的回归模型,预测不同气候区的温度曲线。
7. 可视化回归结果
quartz() par(mfrow=c(2,3)) plot(betaestList[[1]]$fd, lwd=2, xlab="Day", ylab="Mean Temperature", main=regions[1],cex.lab=2,cex.axis=2) for (j in 2:p) plot(betaestList[[j]]$fd, lwd=2, xlab="Day", ylab="Difference", main=regions[j],cex.lab=2,cex.axis=2)将回归结果进行可视化,其中显示了总体加拿大和各个气候区的温度变化。
8. 检验气候区对温度的影响
F.res = Fperm.fd(temp36fd, regionList, betaList) names(F.res) quartz() par(mfrow=c(1,1),mar = c(8, 8, 4, 2)) with(F.res,{ q = 0.95 ylims = c(min(c(Fvals, qval, qvals.pts)), max(c(Fobs, qval))) plot(argvals, Fvals, type = "l", ylim = ylims, col = 1, lwd = 2, xlab = "day", ylab = "F-statistic", cex.lab=1.5, cex.axis=1.5) lines(argvals, qvals.pts, lty = 3, col = 1, lwd = 2) abline(h = qval, lty = 2, col = 1, lwd = 2) legendstr = c("Observed Statistic", paste("pointwise", 1 - q, "critical value"), paste("maximum", 1 - q, "critical value")) legend(argvals[1], 1.4, legend = legendstr, col = c(1, 1, 1), lty = c(1, 3, 2), lwd = c(2, 2, 2)) })利用功能数据的F检验
Fperm.fd来评估气候区对温度的显著影响,并将结果可视化。这段代码通过功能数据分析来研究不同气候区对温度的影响,展示了功能线性回归以及假设检验的结果。