CODEWITHSUNDEEP
r
Trevor Bye
Trevor Bye

Reputation: 708

Format x-axis as Dates in Plot

I'm working with tsoutliers to plot the outliers from a time series, but I'm having trouble getting the dates in the x-axis of the plot to display as actual dates. Right now they are displayed as what looks to be some type of Julian date; 16816, 16816.2, etc. I've researched around and haven't found much on this, but I'm wondering if there is a way to use ggplot() with a ts object, because I know ggplot2 automatically handles dates very well.

Here is the code I'm working with, I just put some quick sample data in for the dataframe, the actual data is coming from a dynamic PowerBI dataframe:

    library(forecast)
    library(tsoutliers)
    library(lubridate)
    library(dplyr)
    library(ggplot2)

    ActualDemand <- c(1000, 250, 500, 3000)
    STRING_DATE <- c("05/13/2017","05/06/2017", "5/20/2017", "05/27/2017")

    dataset <- data.frame(ActualDemand, STRING_DATE)

    #convert from string to date
    dataset$STRING_DATE = lubridate::mdy(dataset$STRING_DATE)
    dataRowCount <- nrow(dataset)

    #find start date
    startDate <- NULL

    for (i in 1:dataRowCount) {
      if (i == 1) {
        startDate <- dataset[i, 2]
        next
      }

      compareDate <- dataset[i,2]

      if (compareDate < startDate) {
        startDate <- compareDate
      } else {
        next
      }
    }

    #transformed set
    datasetSorted <- dataset[order(dataset$STRING_DATE),]

    #build time series, run outlier detection
    ts <- ts(datasetSorted$ActualDemand, frequency = 52, start = startDate)
    series.outliers <- tso(ts)
    plot(series.outliers)

Here is what the output looks like:

enter image description here

Upvotes: 0

Views: 356

Answers (1)

Lstat
Lstat

Reputation: 1460

You can overwrite xaxis values as:

plot(series.outliers, 
  args.x.axis = list(at = time(series.outliers$y), labels=datasetSorted$STRING_DATE) )

Upvotes: 1

Related Questions