Ability to Aggregate Days of Week over X timeframe about grafana-hourly-heatmap-panel

[pconwell]

Just for future proofing (in case the link above becomes dead), here is an example of the type of "aggregated" heatmap I am referencing:

The summary bars on the top/right would be nice - but not critical. This particular chart was generated in R with the following code (yes, I realize R and d3/js have zero overlap):

plotTheHeat <- function(dat) {
  
  plot_w = 0.85
  plot_h = 0.75
  
  
  dat$week <- weekdays(as.Date(dat$ticket_date, format="%m/%d/%Y"), abbreviate=TRUE)
  dat$week <- factor(dat$week, levels=c("Sun", "Mon", "Tue", "Wed", "Thu", "Fri", "Sat"))
  dat$hour <- format(as.POSIXct(dat$ticket_date, format="%m/%d/%Y %H:%M"), "%H")
  dat$hour <- factor(dat$hour, levels=c('00', '01', '02', '03', '04', '05', '06', '07', '08', '09', '10', '11', '12', '13', '14', '15', '16', '17', '18', '19', '20', '21', '22', '23'))
  mat <- table(dat$hour, dat$week)
  
  colors <- colorRampPalette(c("#adddd1", "#3e98af", "#375980"))
  
  
  
  figs <- rbind(c(0.0, plot_w, plot_h, 1.0), # top
                c(plot_w, 1.0, 0.0, plot_h), # right
                c(0.0, plot_w, 0.0, plot_h)  # main
  )
  colnames(figs) <- c("W", "E", "S", "N")
  rownames(figs) <- c("top", "right", "main")
  
  screenIDs <- split.screen(figs)
  names(screenIDs) <- rownames(figs)
  
  screen(screenIDs["top"])
  par(mar=c(0,0,0,plot_w))
  cols <- colors(20)[cut(colSums(mat),20)]
  xs   <- seq_len(ncol(mat))
  ys   <- colSums(mat)
  plot.new()
  plot.window(xlim=c(0, ncol(mat)), ylim=c(0.0, max(ys)*2.5))
  rect(xs-0.95, 0.0, xs-0.05, ys+0.0, col=cols, border="white", lwd=0.5)
  text(xs-0.4, ys, round(ys), pos=3, offset=1.0, srt=90, cex=0.75, font=2)
  
  screen(screenIDs["right"])
  par(mar=c(0,1-plot_w,1-plot_h,0))
  cols <- colors(20)[cut(rowSums(mat),20)]
  xs   <- rowSums(mat)
  ys   <- seq_len(nrow(mat))
  plot.new()
  plot.window(xlim=c(0, max(xs)*2.5), ylim=c(nrow(mat), 0))
  rect(0, ys-0.9, xs, ys-0.1, col=cols, border="white", lwd=0.75)
  text(0, ys-0.5, round(xs), pos=4, cex=0.75, col="black", font=2)
  
  screen(screenIDs["main"])
  par(mar=c(0,0,1-plot_h,plot_h))
  cols <- colors(20)[cut(mat,20)]
  xs   <- col(mat)
  ys   <- row(mat)
  plot.new()
  plot.window(xlim=c(0, max(xs)), ylim=rev(c(0, max(ys))))
  rect(xs-1, ys-1, xs, ys, col=cols, border="white", lwd=0.5)
  text(xs-0.5, ys-0.5, round(mat), cex=0.8, col="white", font=0)
  text(xs[1,]-0.5, 0, colnames(mat), pos=3, cex=0.8, xpd=TRUE)
  text(ncol(mat), ys[,1]-0.5, rownames(mat), pos=4, cex=0.8, xpd=TRUE)
  close.screen(screenIDs)

[pconwell]

This is the type of chart I'm looking at: http://bl.ocks.org/ganezasan/dfe585847d65d0742ca7d0d1913d50e1

It's been a long time since I've touched d.3, but I may try to take a look at putting something together if I find time.

[marcusolsson]

Interesting! So you're essentially transposing the graph and aggregrating into 7 buckets for every day of the week. I think it would be within the scope of the plugin, though I won't be able to look at this myself for a while. If you or anyone else wants to take a shot at it, I'd be happy to review it.

[pconwell]

@marcusolsson Yes! Essentially 168 buckets total (7 days x 24 hours each). Or basically take the current graph and "loop" the 8th day's data back on top of the 1st day, the 9th day's data on top of the 2nd day, etc. In other words, every N%7+X day (where X is 1 - 7 for Sun through Sat) gets aggregated.

I don't have a lot of time in the next month or two either, but after that I should. BUT even then, I'm not super familiar with d3/js, but I'll see if I can work on it.

Thanks!

[dmamalis]

Probably the most missing feature in Grafana! The main reason I go through such plugins every now and then. Awesome work by the way ;-) Full support on this issue!