CODEWITHSUNDEEP
rggplot2geom-tile
Bastounator
Bastounator

Reputation: 11

How to make Heatmap with log y axis

I'm trying to represent particle size distribution with geom_tile, like this plot : [enter image description here] (https://i.sstatic.net/zOGn9uT5.png)

My data represent concentration of particule for each bin of size and for each time. There is 137 bins of size per second.

So my df has 3 columns and 82337 rows. The first column is time from 10:45:00 to 10:56:00, then the numeric value of concentration per time (with value equal to 0 sometimes), and the numeric value of bin_column which represent bin size from 15 to 1375 per second. Here an example (I removed some lines)

Time Bin_column Concentration
10:45:00 15 1.583952e+04
10:45:00 25 5.800000e-01
10:45:00 1375 1.559140e+00
10:45:01 15 0.000000e+00

Here is my code :

library(ncdf4)
library(RNetCDF)
library(ggplot2)
library(data.table)
library(dplyr)
library(scales)
library(patchwork)
library(ggpubr)
library(cowplot)
library(tidyr)
library(purrr)
library(chron)
library(scico)

files <- c(
  "base201204090056_2DS.nc",
  "base201204093146_2DS.nc",
  "base201204102044_2DS.nc",
  "base201204105249_2DS.nc",
  "base201204110956_2DS.nc",
  "base201204130416_2DS.nc",
  "base201204134204_2DS.nc",
  "base201204141637_2DS.nc",
  "base201204144715_2DS.nc",
  "base201204151055_2DS.nc"
)


diameter_h<- seconds_h <- bins<-mvd_all<-numeric()
concentration_h<-size_distrib_h<-matrix()

for (file in files) {
  nc_file <- nc_open(file)
  diameter_h<-c(diameter_h, ncvar_get(nc_file, nc_file$var[[294]]))
  size_distrib_h<-c(size_distrib_h, ncvar_get(nc_file, nc_file$var[[274]]))
  mvd_all<-c(mvd_all, ncvar_get(nc_file, nc_file$var[[285]]))
  seconds_h<-c(seconds_h, ncvar_get(nc_file, nc_file$var[[298]]))
  concentration_h<-c(concentration_h, ncvar_get(nc_file, nc_file$var[[280]]))
  bins<-c(bins, ncvar_get(nc_file, nc_file$var[[272]]))
  nc_close(nc_file)
}

bins<-as.data.frame(bins)
mvd_all<-as.data.frame(mvd_all)
seconds_h<-as.data.frame(seconds_h)
diameter_h<-as.data.frame(diameter_h)
concentration_h<-as.data.frame(concentration_h)
size_distrib_h<-as.data.frame(size_distrib_h)

[...]                 


ggplot(df, aes(x = time, y =bin_column, fill = concentration_h)) +
  geom_tile()+
  scale_fill_scico(palette = 'batlow', trans = "log", labels = trans_format("log10", math_format(10^.x)), breaks = c(0.1,1,10,100,1000,10000), limits=c(0.1,10000))+
  geom_vline(xintercept = as.numeric(time_1048), color = "red", linetype = "dashed", size = 1)+
  geom_vline(xintercept = as.numeric(time_1052), color = "red", linetype = "dashed", size = 1)+
  theme_bw() +
  labs(x = "Time", y = "Max Diameter [μm]", fill = expression("Concentration " *" (#/L/"*mu*"m)")) +
  scale_y_continuous(trans="log10",breaks = c(10,100,1000), limits = c(10,1000), labels = trans_format("log10", math_format(10^.x)))+
  scale_x_datetime(breaks = date_breaks("2 min"), date_labels = "%H:%M")+
  theme(axis.text.x = element_text(angle = 45, hjust = 1), 
        plot.margin = unit(c(1, 1, 1, 1), "cm")) +
  theme(axis.title.x = element_text(size = 18),  
        axis.title.y = element_text(size = 18),
        axis.text.x = element_text(size = 15),   
        axis.text.y = element_text(size = 15))

And the plot : enter image description here

Do you know why my data flattened like that ?

Upvotes: 1

Views: 114

Answers (1)

stefan
stefan

Reputation: 125897

As already mentioned in the comments the issue is due to the log scale. Instead I would suggest to use a geom_rect which however requires some manual computation.

First, let's reproduce your issue using some fake random example data:

library(ggplot2)
library(dplyr, warn = FALSE)

set.seed(123)

dat <- expand.grid(
  x = seq(10),
  y = seq(10, 1000, length.out = 100)
)
dat$fill <- rnorm(nrow(dat))

dat |>
  ggplot(aes(x, y, fill = fill)) +
  geom_tile() +
  scale_y_log10()

While I also tried with setting the height in geom_tile as suggested by @GregorThomas, this does not work well. Instead a geom_rect allows for more flexibility:

dat |>
  dplyr::mutate(
    ylead = abs(y - dplyr::lead(y)),
    ylag = abs(y - dplyr::lag(y)),
    ymax = y + .5 * dplyr::coalesce(ylead, ylag),
    ymin = y - .5 * dplyr::coalesce(ylag, ylead),
    # Restrict lower bound to avoid the values from getting dropped
    ymin = if_else(ymin <= 0, 1, ymin),
    .by = x
  ) |>
  ggplot(aes(fill = fill)) +
  geom_rect(
    aes(
      xmin = x - .5, xmax = x + .5,
      ymin = ymin, ymax = ymax
    )
  ) +
  scale_y_log10(
    breaks = c(10, 100, 1000),
    labels = scales::label_log()
  )

Upvotes: 0

Related Questions