R/plot.ftsf.R

In ftsa: Functional Time Series Analysis

plot.ftsf <- function (x, plot.type = c("function", "components", "variance"),
components, xlab1 = fit$y$xname, ylab1 = "Basis function",
xlab2 = "Time", ylab2 = "Coefficient", mean.lab = "Mean",
level.lab = "Level", main.title = "Main effects", interaction.title = "Interaction",
vcol = 1:3, shadecols = 7, fcol = 4, basiscol = 1, coeffcol = 1,
outlier.col = 2, outlier.pch = 19, outlier.cex = 0.5, ...)
{
    plot.type <- match.arg(plot.type)
    if (plot.type == "function")
    plot(x$mean, ...)
    else if (plot.type == "variance") {
        xx <- x[[1]]$x
        xname <- x[[1]]$xname
        plot(x[[1]]$x, rowMeans(x$var$model), type = "l", xlab = xname,
        ylab = "Variance", col = vcol[1])
        abline(0, 0, lty = 2)
        lines(xx, x$var$mean, col = vcol[2])
        lines(xx, x$var$error, col = vcol[3])
    }
    else {
        fit <- x$model
        oldpar <- par(no.readonly = TRUE)
        mean <- is.element("mean", colnames(fit$basis))
        level <- is.element("level", colnames(fit$basis))
        m <- mean + level
        order <- ncol(fit$basis) - m
        if (missing(components))
        components <- order
        n <- components + (mean | level)
        par(mfcol = c(2, n))
        if (mean)
        plot(fit$y$x, fit$basis[, "mean"], type = "l", lty = 1,
        xlab = xlab1, ylab = mean.lab, main = main.title,
        col = basiscol, ...)
        if (level)
        plot.ts(fit$coeff[, "level"], xlab = xlab2, ylab = level.lab,
        main = ifelse(mean, "", main.title), col = coeffcol,
        ...)
        if (m == 1)
        plot(0, 0, type = "n", xaxt = "n", yaxt = "n", bty = "n",
        xlab = "", ylab = "")
        if (components > 0) {
            for (i in 1:components) {
                yl1 <- ifelse(components > 1, paste(ylab1, i), ylab1)
                yl2 <- ifelse(components > 1, paste(ylab2, i), ylab2)
                plot(fit$y$x, fit$basis[, m + i], type = "l",
                lty = 1, xlab = xlab1, ylab = yl1, col = basiscol,
                ...)
                if (i == 1)
                title(interaction.title)
                plot(x$coeff[[i + m]], type = "n", xlab = xlab2,
                ylab = yl2, main = "", shadecols = shadecols,
                fcol = fcol, col = coeffcol, ...)
                lines(x$coeff[[i + m]]$x, col = "gray")
                junk <- x$coeff[[i + m]]$x
                junk[fit$weights < 1e-06] <- NA
                lines(junk, col = coeffcol)
                lines(x$coeff[[i + m]]$model$fitted, col = fcol)
                if (!is.null(fit$weights))
                points(time(fit$coeff)[fit$weights < 0.1],
                fit$coeff[fit$weights < 0.1, i + m], pch = outlier.pch,
                col = outlier.col, cex = outlier.cex)
            }
        }
        par(oldpar)
    }
}