Plot many line charts on top of gantt chart, quickly/scalably

Question

Objective Chart: (Made with Photoshop)

Monitor a janitor's presence in and out of a hall. Blue is Janitor 1, Brown is Janitor 2. Line represents number of guests present in hall (scale is relative to itself i.e. ymin=0, ymax=max.visitor.value, no explicit axes). Colour represents (cum. janitor time)/(cum. # of guests), i.e. if mess piles up, line starts to turn red.

Where I am: (R with plotrix::gantt.chart)

Question

I am looking to achieve the objective (the addition of gradient line per day) completely with R. I am not sure how to proceed - I'm thinking of either: 1) adding them with lower lever graphics and manually arranging them, since plotrix is based on base graphics or 2) move away from plotrix and use a grid based package, which may be more readable/higher level(?) than the former.

Problems

I tried looking for a grid based package by looking at a SO thread discussing various ways to make gantts. I tried using plotly but it didn't seem to support 'multiple time chunks' on the same 'task'. I wanted to take a pause before committing more time to other packages and possibly running into dead ends (considering the addition of all the line plots) while being ignorant of superior methods.

As a last resort, I may consider creating the line graphs one by one and stitching them to the gantt chart outside of R.

Data Sample

Gantt chart data (random sample, only Monday and Tuesday for brevity)

   janitor.type weekday    dummy.start.time      dummy.end.time
        <dbl>   <chr>              <dttm>              <dttm>
 1          1  Monday 1970-01-01 18:01:20 1970-01-01 18:06:50
 2          1  Monday 1970-01-01 18:08:10 1970-01-01 18:11:52
 3          1  Monday 1970-01-01 17:22:00 1970-01-01 17:23:00
 4          1  Monday 1970-01-01 11:39:40 1970-01-01 11:41:58
 5          2  Monday 1970-01-01 19:35:40 1970-01-01 19:40:40
 6          1  Monday 1970-01-01 15:23:00 1970-01-01 15:24:12
 7          1  Monday 1970-01-01 11:54:50 1970-01-01 12:00:20
 8          1 Tuesday 1970-01-01 17:23:00 1970-01-01 18:18:18
 9          2 Tuesday 1970-01-01 19:25:00 1970-01-01 19:39:18
10          1 Tuesday 1970-01-01 16:40:10 1970-01-01 17:09:10
11          1 Tuesday 1970-01-01 14:16:50 1970-01-01 14:19:38
12          2 Tuesday 1970-01-01 09:27:00 1970-01-01 09:30:30
13          1 Tuesday 1970-01-01 14:08:40 1970-01-01 14:13:40
14          1 Tuesday 1970-01-01 11:12:40 1970-01-01 11:13:40

> dput(gantt)

structure(list(asset.type = c(1, 1, 1, 1, 2, 1, 1, 1, 1, 1, 1, 
1, 1, 1), weekday = c("Monday", "Monday", "Monday", "Monday", 
"Monday", "Monday", "Monday", "Tuesday", "Tuesday", "Tuesday", 
"Tuesday", "Tuesday", "Tuesday", "Tuesday"), dummy.start.time = structure(c(82880, 
83290, 80520, 59980, 88540, 73380, 60890, 80580, 87900, 78010, 
69410, 52020, 68920, 58360), class = c("POSIXct", "POSIXt"), tzone = ""), 
    dummy.end.time = structure(c(83210, 83512, 80580, 60118, 
    88840, 73452, 61220, 83898, 88758, 79750, 69578, 52230, 69220, 
    58420), class = c("POSIXct", "POSIXt"), tzone = "")), row.names = c(NA, 
-14L), vars = "weekday", drop = TRUE, .Names = c("asset.type", 
"weekday", "dummy.start.time", "dummy.end.time"), indices = list(
    0:6, 7:13), group_sizes = c(7L, 7L), biggest_group_size = 7L, labels = structure(list(
    weekday = c("Monday", "Tuesday")), row.names = c(NA, -2L), class = "data.frame", vars = "weekday", drop = TRUE, .Names = "weekday"), class = c("grouped_df", 
"tbl_df", "tbl", "data.frame"))

Visitors

Monday          Tuesday 
9:00:00 AM  138 9:00:00 AM  153
10:00:00 AM 251 10:00:00 AM 299
11:00:00 AM 432 11:00:00 AM 479
12:00:00 PM 560 12:00:00 PM 453
1:00:00 PM  555 1:00:00 PM  535
2:00:00 PM  475 2:00:00 PM  383
3:00:00 PM  448 3:00:00 PM  416
4:00:00 PM  469 4:00:00 PM  417
5:00:00 PM  459 5:00:00 PM  519
6:00:00 PM  403 6:00:00 PM  384
7:00:00 PM  290 7:00:00 PM  278
8:00:00 PM  120 8:00:00 PM  116
9:00:00 PM  29  9:00:00 PM  34

dput(visitors)

structure(list(weekday = c("Monday", "Monday", "Monday", "Monday", 
"Monday", "Monday", "Monday", "Monday", "Monday", "Monday", "Monday", 
"Monday", "Monday", "Tuesday", "Tuesday", "Tuesday", "Tuesday", 
"Tuesday", "Tuesday", "Tuesday", "Tuesday", "Tuesday", "Tuesday", 
"Tuesday", "Tuesday", "Tuesday"), time = structure(c(50400, 54000, 
57600, 61200, 64800, 68400, 72000, 75600, 79200, 82800, 86400, 
90000, 93600, 50400, 54000, 57600, 61200, 64800, 68400, 72000, 
75600, 79200, 82800, 86400, 90000, 93600), class = c("POSIXct", 
"POSIXt"), tzone = ""), count = c(138L, 251L, 432L, 560L, 555L, 
475L, 448L, 469L, 459L, 403L, 290L, 120L, 29L, 153L, 299L, 479L, 
453L, 535L, 383L, 416L, 417L, 519L, 384L, 278L, 116L, 34L)), .Names = c("weekday", 
"time", "count"), row.names = c(NA, -26L), class = "data.frame")

plotrix::gantt.chart code

# Set up variables for gantt chart

labels <- gantt$weekday
starts <- gantt$dummy.start.time
ends <- gantt$dummy.end.time
priorities <- as.numeric(gantt$asset.type)

Ymd.format <- "%Y/%m/%d %H:%M:%S"

# Feed variables to chart parameters

gantt.info <- list(
  labels = labels,
  starts = starts,
  ends = ends,
  priorities = priorities
)

# Define chart intervals

hours <- seq(
  as.POSIXct("1970/01/01 09:00:00", format = Ymd.format),
  as.POSIXct("1970/01/01 21:00:00", format = Ymd.format),
  by = "hour"
)

# Define labels for vgridlab

hourslab <- format(hours, format = "%H")

# Define vertical gridline

vgridpos <- as.POSIXct(hours, format = Ymd.format)
vgridlab <- hourslab

# Optional coloring

colfunc <- colorRampPalette(c("#00bff3", "#362f2d"))

# Define timeframe on x axis

timeframe <-
  as.POSIXct(c("1970/01/01 09:00:00", "1970/01/01 21:00:00"), format = Ymd.format)

# Create the chart

main = ""

test <- gantt.chart(
  gantt.info,
  taskcolors = colfunc(2),
  xlim = timeframe,
  main = main,
  priority.legend = F,
  vgridpos = vgridpos,
  vgridlab = vgridlab,
  hgrid = TRUE,
  half.height = 0.125,
  time.axis = 1
)

Please `dput` your gantt chart and visitor data so we can access them and provide your current `plotrix::gantt.chart` code. I'm inclined to say this can be done in ggplot2. — Djork, Oct 05 '17 at 23:27

Djork · Accepted Answer · 2017-10-06T04:26:03.243

I used ggplot2. To be able to use 2 different color settings -- one for gantt asset type and the other visitor count, I chose two geom types, one able to take fill color: geom_rect, the other line color: geom_line.

I use grid faceting to be able to separate the weekdays as gantt and visitors do not have a common y-axis.

Color and fill palettes can be set either by scale_colour_gradient (continuous) or scale_fill_manual (discrete), or to take advantage of color brewer palettes scale_colour_distiller (continuous) and scale_fill_brewer (discrete). I used an example from each below to replicate color scheme from the given gantt plot.

# factor weekdays and set levels to correct order
gantt$weekday <- factor(gantt$weekday, levels=c("Monday", "Tuesday", "Wednesday", "Thursday", "Friday", "Saturday", "Sunday"))
visitors$weekday <- factor(visitors$weekday, levels=c("Monday", "Tuesday", "Wednesday", "Thursday", "Friday", "Saturday", "Sunday"))

# factor gantt$asset.type as this represent categorical not continuous values 
gantt$asset.type <- factor(gantt$asset.type)


# gantt chart
library(ggplot2)
ggplot() +
  geom_line(data=visitors, aes(x=time, y=count, colour=count)) +
  geom_rect(data=gantt, aes(xmin=dummy.start.time, xmax=dummy.end.time, ymin=-100, ymax=0, fill=asset.type)) +
  facet_grid(weekday~.) +
  scale_colour_distiller("Visitor Count", type="div", palette="RdYlGn", direction = -1) + 
  scale_fill_manual("Asset Type", values=c("dodger blue", "black")) +  
  theme_bw() +
  theme(axis.title = element_blank())

Note: You can adjust the thickness of the Asset bars by adjusting the ymin and ymax of geom_rect. I had ymax=0 and ymin taking a negative value so it is plotted below the Visitor Count y-range.

I should add that you can also make gantt charts in ggplot2 using geom_segment, but a particular issue with your plot requires two different color scales, and geom_segment only takes colour aesthetics like geom_line.

Here is ggplot2 code for just creating gantt charts:

ggplot(gantt) +
  geom_segment(aes(x=dummy.start.time, xend=dummy.end.time, y=weekday, yend=weekday, colour=asset.type), size=10) +
  scale_colour_manual("Asset Type", values=c("dodger blue", "black")) +  
  theme_bw() +
  theme(axis.title = element_blank())

Plot many line charts on top of gantt chart, quickly/scalably

1 Answers1