pkg/R/plot.R

   1 #' Plot curves
   2 #'
   3 #' Plot a range of curves in data.
   4 #'
   5 #' @inheritParams computeError
   6 #' @param indices Range of indices (integers or dates)
   7 #'
   8 #' @export
   9 plotCurves <- function(data, indices=seq_len(data$getSize()))
  10 {
  11     series = data$getSeries(indices)
  12     yrange = quantile(series, probs=c(0.025,0.975), na.rm=TRUE)
  13     par(mar=c(4.7,5,1,1), cex.axis=1.5, cex.lab=1.5)
  14     matplot(series, type="l", ylim=yrange, xlab="Time (hours)", ylab="PM10")
  15 }
  16
  17 #' Plot error
  18 #'
  19 #' Draw error graphs, potentially from several runs of \code{computeForecast()}.
  20 #'
  21 #' @param err Error as returned by \code{computeError()}
  22 #' @param cols Colors for each error (default: 1,2,3,...)
  23 #'
  24 #' @seealso \code{\link{plotCurves}}, \code{\link{plotPredReal}},
  25 #'   \code{\link{plotSimils}}, \code{\link{plotFbox}}, \code{\link{computeFilaments}},
  26 #'   \code{\link{plotFilamentsBox}}, \code{\link{plotRelVar}}
  27 #'
  28 #' @export
  29 plotError <- function(err, cols=seq_along(err), agg="day")
  30 {
  31     if (!is.null(err$abs))
  32         err = list(err)
  33     par(mfrow=c(2,2), mar=c(4.7,5,1,1), cex.axis=1.5, cex.lab=1.5)
  34     L = length(err)
  35
  36     yrange = range( sapply(1:L, function(i) err[[i]]$abs$day), na.rm=TRUE )
  37     matplot(sapply( seq_len(L), function(i) err[[i]]$abs$day ), type="l",
  38         xlab="Time (hours)", ylab="Mean |y - y_hat|", ylim=yrange, col=cols, lwd=2, lty=1)
  39
  40     agg_curves <- sapply( seq_len(L), function(i) {
  41         curve <- err[[i]]$abs$indices
  42         delta <- if (agg=="day") 1 else if (agg=="week") 7 else if (agg=="month") 30
  43         vapply( seq(1,length(curve),delta), function(i) {
  44             mean(curve[i:(i+delta-1)], na.rm=TRUE)
  45         }, vector("double",1), USE.NAMES=FALSE )
  46     })
  47     yrange = range(agg_curves, na.rm=TRUE)
  48     matplot(agg_curves, type="l", xlab=paste("Time (",agg,"s)", sep=""),
  49         ylab="Mean |y - y_hat|", ylim=yrange, col=cols, lwd=2, lty=1)
  50
  51     yrange = range( sapply(1:L, function(i) err[[i]]$MAPE$day), na.rm=TRUE )
  52     matplot(sapply( seq_len(L), function(i) err[[i]]$MAPE$day ), type="l",
  53         xlab="Time (hours)", ylab="Mean MAPE", ylim=yrange, col=cols, lwd=2, lty=1)
  54
  55     agg_curves <- sapply( seq_len(L), function(i) {
  56         curve <- err[[i]]$MAPE$indices
  57         delta <- if (agg=="day") 1 else if (agg=="week") 7 else if (agg=="month") 30
  58         vapply( seq(1,length(curve),delta), function(i) {
  59             mean(curve[i:(i+delta-1)], na.rm=TRUE)
  60         }, vector("double",1), USE.NAMES=FALSE )
  61     })
  62     yrange = range(agg_curves, na.rm=TRUE)
  63     matplot(agg_curves, type="l", xlab=paste("Time (",agg,"s)", sep=""),
  64         ylab="Mean MAPE", ylim=yrange, col=cols, lwd=2, lty=1)
  65 }
  66
  67 #' Plot measured / predicted
  68 #'
  69 #' Plot measured curve (in black) and predicted curve (in blue).
  70 #'
  71 #' @inheritParams computeError
  72 #' @param index Index in forecasts (integer or date)
  73 #'
  74 #' @export
  75 plotPredReal <- function(data, pred, index)
  76 {
  77     prediction = pred$getForecast(index)
  78     measure = data$getSerie( pred$getIndexInData(index) )[1:length(pred$getForecast(1))]
  79
  80     # Remove the common part, where prediction == measure
  81     dot_mark <- ifelse(prediction[1]==measure[1],
  82         which.max(seq_along(prediction)[prediction==measure]), 0)
  83     prediction = prediction[(dot_mark+1):length(prediction)]
  84     measure = measure[(dot_mark+1):length(measure)]
  85
  86     yrange = range(measure, prediction)
  87     par(mar=c(4.7,5,1,1), cex.axis=1.5, cex.lab=1.5, lwd=3)
  88     plot(measure, type="l", ylim=yrange, xlab="Time (hours)", ylab="PM10")
  89     par(new=TRUE)
  90     plot(prediction, type="l", col="#0000FF", ylim=yrange, xlab="", ylab="")
  91 }
  92
  93 #' Plot similarities
  94 #'
  95 #' Plot histogram of similarities (weights), for 'Neighbors' method.
  96 #'
  97 #' @inheritParams computeError
  98 #' @param index Index in forecasts (integer or date)
  99 #'
 100 #' @export
 101 plotSimils <- function(pred, index)
 102 {
 103     weights = pred$getParams(index)$weights
 104     if (is.null(weights))
 105         stop("plotSimils only works on 'Neighbors' forecasts")
 106     par(mfrow=c(1,2), mar=c(4.7,5,1,1), cex.axis=1.5, cex.lab=1.5)
 107     small_weights = weights[ weights < 1/length(weights) ]
 108     large_weights = weights[ weights >= 1/length(weights) ]
 109     hist(small_weights, nclass=25, main="", xlab="Weight < 1/N", ylab="Count")
 110     hist(large_weights, nclass=25, main="", xlab="Weight >= 1/N", ylab="Count")
 111 }
 112
 113 #' Functional boxplot
 114 #'
 115 #' Draw the functional boxplot on the left, and bivariate plot on the right.
 116 #'
 117 #' @inheritParams computeError
 118 #' @inheritParams plotCurves
 119 #'
 120 #' @export
 121 plotFbox <- function(data, indices=seq_len(data$getSize()))
 122 {
 123     if (!requireNamespace("rainbow", quietly=TRUE))
 124         stop("Functional boxplot requires the rainbow package")
 125
 126     series_matrix = data$getSeries(indices)
 127     # Remove series with NAs
 128     no_NAs_indices = sapply( 1:ncol(series_matrix),
 129         function(i) all(!is.na(series_matrix[,i])) )
 130     series_matrix = series_matrix[,no_NAs_indices]
 131
 132     series_fds = rainbow::fds(seq_len(nrow(series_matrix)), series_matrix)
 133     par(mar=c(4.7,5,1,1), cex.axis=1.5, cex.lab=1.5)
 134     rainbow::fboxplot(series_fds, "functional", "hdr", xlab="Time (hours)", ylab="PM10",
 135         plotlegend=FALSE, lwd=2)
 136     rainbow::fboxplot(series_fds, "bivariate", "hdr", plotlegend=FALSE)
 137 }
 138
 139 #' Compute filaments
 140 #'
 141 #' Obtain similar days in the past, and (optionally) plot them -- as black as distances
 142 #' are small.
 143 #'
 144 #' @inheritParams computeError
 145 #' @param index Index in forecast (integer or date)
 146 #' @param limit Number of neighbors to consider
 147 #' @param plot Should the result be plotted?
 148 #'
 149 #' @return A list with
 150 #' \itemize{
 151 #'   \item index : index of the current serie ('today')
 152 #'   \item neighb_indices : indices of its neighbors
 153 #'   \item colors : colors of neighbors curves (shades of gray)
 154 #' }
 155 #'
 156 #' @export
 157 computeFilaments <- function(data, pred, index, limit=60, plot=TRUE)
 158 {
 159     weights <- pred$getParams(index)$weights
 160     if (is.null(weights) || is.na(pred$getParams(index)$weights[1]))
 161         stop("computeFilaments requires a serie without NAs")
 162
 163     nn <- min(limit, length(weights))
 164     sorted_dists = sort(-log(weights), index.return=TRUE)
 165     # Compute colors for each neighbor (from darkest to lightest), if weights differ
 166     if ( any( weights != weights[1] ) )
 167     {
 168         min_dist = min(sorted_dists$x[1:nn])
 169         max_dist = max(sorted_dists$x[1:nn])
 170         color_values = floor(19.5*(sorted_dists$x[1:nn]-min_dist)/(max_dist-min_dist)) + 1
 171         colors = gray.colors(20,0.1,0.9)[color_values] #TODO: 20 == magic number
 172     }
 173     else
 174         colors <- rep(colors()[17], length(weights))
 175
 176     if (plot)
 177     {
 178         # Complete series with (past and present) tomorrows
 179         ref_serie = c( data$getCenteredSerie( pred$getIndexInData(index)-1 ),
 180             data$getCenteredSerie( pred$getIndexInData(index) ) )
 181         centered_series = rbind(
 182             data$getCenteredSeries( pred$getParams(index)$indices-1 ),
 183             data$getCenteredSeries( pred$getParams(index)$indices ) )
 184         yrange = range( ref_serie,
 185             quantile(centered_series, probs=c(0.025,0.975), na.rm=TRUE) )
 186         par(mar=c(4.7,5,1,1), cex.axis=1.5, cex.lab=1.5, lwd=2)
 187         for (i in nn:1)
 188         {
 189             plot(centered_series[,sorted_dists$ix[i]], ylim=yrange, type="l", col=colors[i],
 190                 xlab=ifelse(i==1,"Time (hours)",""), ylab=ifelse(i==1,"Centered PM10",""))
 191             par(new=TRUE)
 192         }
 193         # Also plot ref curve, in red
 194         plot(ref_serie, ylim=yrange, type="l", col="#FF0000", xlab="", ylab="")
 195         dot_mark <- 0.5 + which.max( pred$getForecast(1) ==
 196             data$getSerie( pred$getIndexInData(1) )[1:length(pred$getForecast(1))] )
 197         abline(v=24+dot_mark, lty=2, col=colors()[56], lwd=1)
 198     }
 199
 200     list(
 201         "index"=pred$getIndexInData(index),
 202         "neighb_indices"=pred$getParams(index)$indices[sorted_dists$ix[1:nn]],
 203         "colors"=colors)
 204 }
 205
 206 #' Functional boxplot on filaments
 207 #'
 208 #' Draw the functional boxplot on filaments obtained by \code{computeFilaments()}.
 209 #'
 210 #' @inheritParams computeError
 211 #' @param fil Output of \code{computeFilaments}
 212 #' @param predict_from First predicted time step
 213 #'
 214 #' @export
 215 plotFilamentsBox = function(data, fil, predict_from)
 216 {
 217     if (!requireNamespace("rainbow", quietly=TRUE))
 218         stop("Functional boxplot requires the rainbow package")
 219
 220     series_matrix = rbind(
 221         data$getSeries(fil$neighb_indices-1), data$getSeries(fil$neighb_indices) )
 222     series_fds = rainbow::fds(seq_len(nrow(series_matrix)), series_matrix)
 223
 224     par(mar=c(4.7,5,1,1), cex.axis=1.5, cex.lab=1.5)
 225     rainbow::fboxplot(series_fds, "functional", "hdr", xlab="Time (hours)", ylab="PM10",
 226         plotlegend=FALSE, lwd=2)
 227
 228     # "Magic": http://stackoverflow.com/questions/13842560/get-xlim-from-a-plot-in-r
 229     usr <- par("usr")
 230     yr <- (usr[4] - usr[3]) / 27
 231     par(new=TRUE)
 232     plot(c(data$getSerie(fil$index-1),data$getSerie(fil$index)), type="l", lwd=2, lty=2,
 233         ylim=c(usr[3] + yr, usr[4] - yr), xlab="", ylab="")
 234     abline(v=24+predict_from-0.5, lty=2, col=colors()[56])
 235 }
 236
 237 #' Plot relative conditional variability / absolute variability
 238 #'
 239 #' Draw the relative conditional variability / absolute variability based on filaments
 240 #' obtained by \code{computeFilaments()}.
 241 #'
 242 #' @inheritParams computeError
 243 #' @inheritParams plotFilamentsBox
 244 #'
 245 #' @export
 246 plotRelVar = function(data, fil, predict_from)
 247 {
 248     ref_var = c( apply(data$getSeries(fil$neighb_indices-1),1,sd),
 249         apply(data$getSeries(fil$neighb_indices),1,sd) )
 250     tdays = .getNoNA2(data, 2, fil$index)
 251     global_var = c(
 252         apply(data$getSeries(tdays-1),1,sd),
 253         apply(data$getSeries(tdays),1,sd) )
 254
 255     yrange = range(ref_var, global_var)
 256     par(mar=c(4.7,5,1,1), cex.axis=1.5, cex.lab=1.5)
 257     plot(ref_var, type="l", col=1, lwd=3, ylim=yrange,
 258         xlab="Time (hours)", ylab="Standard deviation")
 259     par(new=TRUE)
 260     plot(global_var, type="l", col=2, lwd=3, ylim=yrange, xlab="", ylab="")
 261     abline(v=24+predict_from-0.5, lty=2, col=colors()[56])
 262 }