Time data in x axis not ordered in R ggplot

Question

I am using the following data frame in R

TIME                 PRICE
2013-01-01 23:55:03 446.6167
2013-01-01 23:55:02 441.0114
2013-01-01 23:54:59 446.7600

I am using function ggplot to plot the data points and label fixed intervals using scale_x_datetime.

library(ggplot2)  # including necessary libraries
lims <- as.POSIXct(strptime(c("2013-01-01 00:00:00","2013-01-01 23:59:00"), format = "%Y-%m-%d %H:%M:%S"))


ggplot(open,aes(x=TIME,y=PRICE))  
   + geom_point(size = 1.0, color="navy") 
   + xlab("Time") 
   + ylab("Price") 
   + ggtitle("time vs Price ") 
   + scale_x_datetime(breaks = date_breaks("200 min"), minor_breaks=date_breaks("15 min"), labels=date_format("%H:%M:%S"),limits=lims)

Despite specifying the limits, the x axis labels are not in order as shown beneath:

Could you post the output of `str(open)` at the bottom of your question? — lmo, Jun 15 '17 at 13:54
Also avoid names to objects such as `open` (since it is a function of `R`) — amonk, Jun 15 '17 at 14:14
This looks like it could be a time zone issue. Are you using the most current version of ggplot2? The most current version defaults to using the time zone attached to the variable, but I remember that whatever the previous default was could lead to problems. — aosmith, Jun 15 '17 at 20:42
A likely duplicate [here](https://stackoverflow.com/questions/36227130/r-as-posixct-timezone-and-scale-x-datetime-issues-in-my-dataset) — aosmith, Jun 15 '17 at 20:43

Pierre Lapointe · Answer 1 · 2017-06-15T14:34:06.267

1

You need to put as.POSIXct(TIME) in you aes like so. I had to change your code a bit to fit your limited 3-point data example.

open <- read.table(text="TIME                 PRICE
'2013-01-01 23:55:03' 446.6167
'2013-01-01 23:55:02' 441.0114
'2013-01-01 23:54:59' 446.7600",header=TRUE, stringsAsFactors=FALSE)

lims <- as.POSIXct(strptime(c("2013-01-01 00:00:00","2013-01-01 23:59:00"),
                   format = "%Y-%m-%d %H:%M:%S"))

ggplot(open,aes(x=as.POSIXct(TIME),y=PRICE)) +  
geom_point(size = 3.0, color="red")+ 
xlab("Time")+ 
ylab("Price")+ 
ggtitle("time vs Price ")+
scale_x_datetime(breaks = date_breaks("360 min"), 
                 minor_breaks=date_breaks("15 min"), 
                 labels=date_format("%H:%M:%S"),limits=lims)

EDIT You should try to use xts to transform your data in a time series object. This will order the time correctly. Then, you fortify it to use with ggplot2.

library(xts)
open <- read.table(text="TIME                 PRICE
'2013-01-01 23:55:03' 446.6167
'2013-01-01 23:55:02' 441.0114
'2013-01-01 23:54:59' 446.7600",header=TRUE, stringsAsFactors=FALSE)
open <- xts(open$PRICE,as.POSIXct(open$TIME))
open <- fortify(open)

lims <- as.POSIXct(strptime(c("2013-01-01 00:00:00","2013-01-01 23:59:00"),
                            format = "%Y-%m-%d %H:%M:%S"))

ggplot(open,aes(x=Index,y=open)) +  
geom_point(size = 3.0, color="green")+ 
xlab("Time")+ 
ylab("Price")+ 
ggtitle("time vs Price ")+
scale_x_datetime(breaks = date_breaks("360 min"), 
                 minor_breaks=date_breaks("15 min"), 
                 labels=date_format("%H:%M:%S"),limits=lims)

edited Jun 15 '17 at 14:34

answered Jun 15 '17 at 13:56

Pierre Lapointe

16,017
2
43
56

@SahityaSridhar Not sure what you are saying. In my example, the x-labels are ordered from midnight to midnight. – Pierre Lapointe Jun 15 '17 at 14:21
In my example the x label ordering doesn't change at all. Could it be because of the difference in the data frame? – Sahitya Sridhar Jun 15 '17 at 14:23
Even when i execute your code i dont get the labels in the right order – Sahitya Sridhar Jun 15 '17 at 14:30
@SahityaSridhar I made an edit which uses `xts`, which is very effective for time series data. It will also order the data correctly. – Pierre Lapointe Jun 15 '17 at 14:37
Nope. Still the same ordering. – Sahitya Sridhar Jun 15 '17 at 14:44
Do you see factors in `str(open)`? If so, you should add `, stringsAsFactors=FALSE` – Pierre Lapointe Jun 15 '17 at 15:05
There aren't any factors in str(open) – Sahitya Sridhar Jun 15 '17 at 15:10

score 0 · Answer 2 · answered Jun 15 '17 at 14:13

I would use lubridate:

library(data.table)
library(lubridate)
library(ggplot2)
time<-seq(ymd_hms('2013-01-01 00:00:01'),ymd_hms('2013-01-02 23:59:59'), by = '1 min')
price<-sample(length(time))
dt<-data.table(time,price)

ggplot(dt,aes(x=time,y=price))  + geom_point(size = 1.0, color="navy") + xlab("Time") + ylab("Price")  + ggtitle("time vs Price ")

resulting in:

Time data in x axis not ordered in R ggplot

2 Answers2