how to save a ggplot output in a pdf format?

Question

I have the following data (this is a head of 10 rows from my data which has 43 days and 24/day hours data):

structure(list(Agency.Station.ID = c("MI270N003.6D", "MI270N004.7D", 
"MI270N005.7D", "MI270N007.3D", "MI270N008.5D", "MI270N003.6D", 
"MI270N004.7D", "MI270N005.7D", "MI270N007.3D", "MI270N008.5D"
), Date = structure(c(16526, 16526, 16526, 16526, 16526, 16526, 
16526, 16526, 16526, 16526), class = "Date"), Time = c("00:00:00", 
"00:00:00", "00:00:00", "00:00:00", "00:00:00", "00:05:00", "00:05:00", 
"00:05:00", "00:05:00", "00:05:00"), Speed = c(59, 34, 46, 61, 
46, 58, 39, 51, 36, 52), Precipitation = c(0, 0, 0, 0, 0, 0, 
0, 0, 0, 0)), row.names = c(NA, 10L), class = "data.frame")

I want to make a pdf file for each "Agency.Station.ID". Each page in the pdf represents a "Date". In each page plot "Time VS Speed and Precipitation), is that possible?

Answer 1

Without your full dataset, it's not possible to show a complete working solution. With that being said... here's a complete working solution with a dummy dataset. First, the dataset:

library(lubridate)
library(ggplot2)
library(tidyr)
library(dplyr)
set.seed(1234)
df1 <- data.frame(
  Agency.Station.ID=rep('Station1',30),
  Date=c(rep(as.Date('2020-01-01'),10), rep(as.Date('2020-02-01'),10), rep(as.Date('2020-03-01'),10)),
  Time=rep(1:10, 3),
  Speed=c(rnorm(10,40,2), rnorm(10,45,2), rnorm(10,56,1)),
  Precipitation=rep(runif(10,0,30), 3)
)

df2 <- data.frame(
  Agency.Station.ID=rep('Station2',30),
  Date=c(rep(as.Date('2020-01-02'),10), rep(as.Date('2020-03-01'),10), rep(as.Date('2020-04-01'),10)),
  Time=rep(1:10, 3),
  Speed=c(rnorm(10,45,11), rnorm(10,35,7), rnorm(10,53,3)),
  Precipitation=rep(runif(10,0,30), 3)
)

df <- rbind(df1, df2)

It's a bunch of nonsense, but it will work for our purposes. I have two stations: "Station1" and "Station2". Each station here has different dates, with one overlapping. I'm not sure if your dataset has all stations with the same dates, but this code should work in either case.

The basic idea is to use the method shown in this answer to make the pdf files. It is setup as written to create one *.pdf file for each value df$Agency.Station.ID, with one page per plot, and one plot for each value of df$Date. I'm plotting the values of df$Speed and df$Precipitation as line geoms vs. df$Time.

Prepare your dataset

First of all, you should adjust your data to be able to create the intended plot (as you indicate in your question). This means that you will need to gather() df$Precipitation and df$Speed into two columns: one for the "Aspect" (either "Precipitation" or "Speed"), and one for the actual "Value". We will also want to ensure df$Agency.Station.ID and df$Date are converted to factors, since we will be stepping through their levels() to create the plots and pdfs:

df <- df %>% gather(Aspect, Value, -Agency.Station.ID, -Date, -Time)
df$Agency.Station.ID <- factor(df$Agency.Station.ID)
df$Date <- factor(df$Date)

The Function

For the function to create the plots, the plan is to have two nested for loops. The outer loop steps through each level in df$Agency.Station.ID and creates a separate pdf file for each. The inner loop steps through each level in df$Date and creates a plot for each:

for(i in levels(df$Agency.Station.ID)) {
  pdf(paste0(i,'.pdf'))

  for(j in levels(df$Date)) {
    subset.data <- df[which(df$Agency.Station.ID==i & df$Date==j),]
    if (nrow(subset.data)!=0) {
      p <- ggplot(data=subset.data, aes(Time, Value)) +
        geom_line(aes(color=Aspect)) +
        labs(title=paste('Station:',i, " Date:",j)) +
        theme_bw()
      invisible(print(p))
    }
  }
  dev.off()
}

You will note that I'm using an if statement to check if the intersection of the station and the date chosen actually has any rows, and if not, no plot is drawn. This is how I'm able to print one page for each date when there are different (or same) dates per station ID. Additionally, note the use of invisible(), which is a nifty function useful for this purpose, where the results of print() don't actually show up in the console, but are still able to be sent to the pdf() graphics device.

If you want additional useability, you can specify a width= and height= for each plot within the print() function. You can also, of course, adjust the aesthetics of the particular plots how you wish. Example of one of the plots below: