Convert date to continuous scale / variable

Question

Is there a way to convert or instruct ggplot to interpret a Dates column as a continuous variable?

My data ( df ) is as follows:

 Location Date Value 56.28,-36.57 2011-01-10 32 56.28,-36.57 2010-02-08 40 52.24,-36.58 2010-03-22 18 52.24,-36.58 2011-06-14 39 52.25,-36.59 2012-04-10 41 52.25,-36.59 2010-04-09 38 

I tried to build the data with the following command:

 g=ggplot(df) + geom_boxplot(aes(factor(Location),Value, col=Date))+ geom_jitter(aes(factor(Location),Value),size=1) + scale_colour_gradient(low='red',high='green') 

But received the following error message:

 Error: Discrete value supplied to continuous scale 

If I convert a Date to Date object (e.g. col=as.Date(Date) ), I get the following error:

 Error in unit(tic_pos.c, "mm") : 'x' and 'units' must have length > 0 

The goal is for the Date column to define the color of the dots, with the earliest dates having color red and later dates having green in the color gradient.

Answer 1

One option is to enclose a date column in as.numeric . However, as Hadley noted in the comments, the trans argument to scale_colour_gradient can take the value date . This has the added benefit (compared to what I previously posted) of a legend displaying date values ​​rather than numeric ones.

Here it is generally:
Also note that I moved the col argument to geom_jitter (not geom_boxplot )

 ggplot(df) + geom_boxplot(aes(factor(Location),Value)) + geom_jitter(aes(factor(Location),Value, col=Date),size=2) + # <~~~ col scale_colour_gradient(trans="date", low="red", high="green") + xlab("Location") 

---

previous answer using as.numeric for comparison

You can wrap a column in as.numeric . Also, I moved the col argument to geom_jitter (not geom_boxplot ).

 ggplot(df) + geom_boxplot(aes(factor(Location),Value))+ geom_jitter(aes(factor(Location),Value, col=as.numeric(Date)),size=2) + scale_colour_gradient(low='red',high='green') + theme(legend.position="none") + xlab("Location") 

Answer 2

You can try to take the min and max of the Date column and display dates on a scale in the range from 0 to 1.

 df$Date=as.POSIXct(df$Date) min=min(df$Date) max=max(df$Date) as.numeric(difftime(df$Date,min,units='days'))/as.numeric(difftime(max,min,units='days')) [1] 0.42426474 0.00000000 0.05298048 0.61992950 1.00000000 0.07570895 

Add this to your data frame and you must be in business.