How to align facet_wrap

Question

I am having a problem aligning my graphics using facet_wrap(). I have multiple years of data but I am subsetting to display only two years. For some unknown reason the upper plot is shifted to the left and the lower plot is shifted to the right (See attached snapshot). My dataset is very large to post here so it can be downloaded from the link below if someone is willing to help me to align the plots. https://login.filesanywhere.com/fs/v.aspx?v=8c6b678a5c61707ab0ae

 Here is the snapshot:

This is what I have tried:

library(writexl)
library(readxl)
library(tidyverse)
library(lubridate)
mydat <- read_excel("all.xlsx", sheet="Sheet1")  

#subset 2 years
start <- 1998
end <- 1999
a <- dplyr::filter(mydat, year %in% start:end)

  ggplot(a,aes(date,Salinity,color=Box)) + 
                  geom_line(size=.8) +
                  theme(legend.position = 'none')  +
       scale_x_date(date_breaks = "1 month",date_labels = "%b",expand=c(0,0.5)) +
                              labs(x="",y= "test") +
                                 facet_wrap(~year,ncol=1) 

I will be subsetting up to 10 years in the future. I am also wondering what's the best to subset multiple years using dplyr or base. Thanks beforehand.

I get the following error after trying your suggestion:

subset 2 years
start <- 1998
end <- 2000

a <- mydat %>%
    dplyr::filter(year %in% start:end) %>% # if you need to subset the years
    mutate(date = as.Date(gsub("\\d{4}", "0000", date)))
     
     
    ggplot(a,aes(date, Salinity,color=Box)) + 
    geom_line(size=.8) + 
    scale_x_date(date_breaks = "1 month",date_labels = "%b",expand=c(0,0.5)) + 
    labs(x="",y= "test") +
    facet_wrap(~year,ncol=1,scales="free_x")

   Error in charToDate(x) : 
  character string is not in a standard unambiguous format

Maybe 'date' doesn't like the zeros? Question: Does it work for you using the dataset from the link?

Answer 1

The plot from the OP appears shifted because the date column in the data contains separate years. Thus, the ggplot function is displaying those dates chronologically along the x-axis, spanning years.

A solution is to change the year in all the dates in the date column to be the same (e.g., "1111"), while the actual year information is retained in the year column. The years in the year column is used to stratify the plot.

# example data
mydat <- data.frame(
    date = as.Date(
        c("1998-10-1", "1998-11-1", "1999-10-1", "1999-11-1", "1999-12-1")), 
    value = 1:5, 
    year = c(1998, 1998, 1999, 1999, 1999))

mydat
#        date value year
#1 1998-10-01     1 1998
#2 1998-11-01     2 1998
#3 1999-10-01     3 1999
#4 1999-11-01     4 1999
#5 1999-12-01     5 1999

# solution by changing the year in all the dates the same
library(ggplot2)

mydat$date <- as.Date(gsub("\\d{4}", "1111", mydat$date))

ggplot(mydat, aes(date, value)) + 
    geom_line(size=.8) + 
    scale_x_date(date_breaks = "1 month",date_labels = "%b",expand=c(0,0.5)) + 
    labs(x="",y= "test") +
    facet_wrap(~year, ncol = 1)

In gsub("\\d{4}", "1111", mydat$date), a regular expression is used for the pattern. It finds four consecutive digits and substitutes them with "1111".

EDIT

If you don't want to make a hard change to the actual date data, you can use pipe operators (%>%) and other functions from the dplyr package:

library(ggplot2)
library(dplyr) # or library(tidyverse) works, too

# subset 2 years
start <- 1998
end <- 1999

mydat %>%
    filter(year %in% start:end) %>% # if you need to subset the years
    mutate(date = as.Date(gsub("\\d{4}", "1111", date))) %>% 
    ggplot(aes(date, value)) + 
    geom_line(size = 0.8) + 
    scale_x_date(date_breaks = "1 month", date_labels = "%b", expand = c(0,0.5)) + 
    labs(x = "", y = "test") +
    facet_wrap(~year, ncol = 1)

UPDATE NOTE

I noticed that the year 0000 (or with just 0) or any year that is 9999+ do not work in the OP's dataset. The year may have to be an integer between 1 and 9998. As long as the arbitrary year is the same throughout the date column, the plot will work as intended.