Cannot set x axis as dates in ggplot2

Question

first time poster, long time lurker on the R questions...

I've finally been stumped for 48 hours, and I'm coming to all of you beat; but hopefully not for long!

I have an irregular time series that I'm trying to plot with ggplot 2. I would like for the breaks and labels to show only for the days where we have data.

The variable holding the date starts off as a factor read in from excel which I convert to class date:

Dataset[,BatchDateCol] <- as.Date(Dataset[,BatchDateCol],format="%m/%d/%y")

(this is part of a larger system, so I can't just read in the data differently)

I create a vector of the labels in the format I want:

Date_Vec <- c(history[,BatchDateCol])
Date_Vec <- format(Date_Vec, "%b %d")

I then aggregate around the days to get the means for each day:

history <- ddply(Dataset, BatchDateCol, function(z) colMeans(z[RawLocation]))

Now I want to plot it, to do that I find the column with the date, and the variable of interest:

ProductionDate <- grep(BatchDateCol,colnames(history))
Location <- grep(GGVar, colnames(history))

This is where the problem starts; I can create my plot just fine like this:

plot2 <- ggplot(history, aes_string(x=history[ProductionDate], y=history[(Location)]))
plot2 + xlab(XAxisName) +ylab(GGVar)+geom_line(aes_string(y=means), linetype="dashed") 
plot2 + geom_point() + geom_smooth(method="lm", formula=y~poly(x,1)) 

but when I try to add dates to the x axis I get errors, no plot, or no breaks+labels. These 2 commands draw the right plot, but with no breaks or labels:

scale_x_continuous(labels=Date_Vec, breaks=c(1:length(history[,BatchDateCol])),
expand=c(.01,0))

&

scale_x_continuous(labels=Date_Vec, breaks=c(1:length(history[,BatchDateCol])),
limits=c(min(as.numeric(history[,BatchDateCol])),
max(as.numeric(history[,BatchDateCol]))))

This command draws the right breaks and labels but no plot(...!)

scale_x_continuous(labels=Date_Vec, 
limits=c(1,length(history[,BatchDateCol])), expand=c(.01,0)) 

and when I plot it with only:

scale_x_continuous(labels=Date_Vec, expand=c(.01,0)) 

it sometimes works, but most of the time I get:

Error in scale_labels.continuous(scale, major) : Breaks and labels are different lengths

If I don't specify labels in scale_x_continuous, I get the numeric form of the dates (days since 1970 or whatever) that I want (though I'm not sure its plotting it in the right spot), but I can't figure out how to modify that either.

Finally I've tried changing scale_x_continuous to scale_x_date:

plot2 + scale_x_date(expand=c(.01,0))

which returns an error: (I've tried placing a few different parameters in the ())

Error : Invalid input: date_trans works with objects of class Date only

I have tried leaving Dataset[,BatchDateCol] as a factor or character vector which also does not work.

SO.... I'm at a total loss and feel unbelievably defeated :(

EDIT

ProductionDate is a quoted variable defined here: QC_Process <- function(Dataset, GGVar, XAxisName="TIME SERIES", BatchDateCol, BatchNum=-1, startdate=NULL, enddate=NULL) {...

So I can't use $ to access the var. Also, I already apply unique right after the var is defined (`history <- unique(history)')

(This is production code for a big R Shiny app, and belongs to my employer, so I can't post to much of it... )

When I use breaks=(history[,ProductionDate]) I get:

Error in as.Date.numeric(value) : 'origin' must be supplied

and When I use 'breaks=(history[ProductionDate])' I get:

Error in is.finite(x) : default method not implemented for type 'list'

print(history[ProductionDate]) returns:

Date.of.Consumption

1 2012-03-24

2 2013-03-11

3 2013-05-10

4 2013-05-11

5 2013-05-13

6 2013-05-16

Answer 1

Got it to work by setting

scale_x_continuous(labels=Date_Vec, Breaks=(as.numeric(history[,ProductionDate])))

Still not sure why it works, and the labels sometimes overlap when I have consecutive dates, but it is progress!

Answer 2

scale_x_continuous(breaks=unique(history$ProductionDate)) should do it (but I can't quite replicate your code, because I don't have the original Dataset, so I haven't been able to test this solution.