CODEWITHSUNDEEP
rggplot2time-series
feder80
feder80

Reputation: 1351

ggplot2, time series and changepoint package

I have the following Dataset (Bu.Dis):

row.names   Date     DIS
    116      1Q1    0.1120
    117      1Q2    0.1104
    118      1Q3    0.3794
    119      1Q4    0.3983
    120      2Q1    0.3175
    121      2Q2    0.2275
    122      2Q3    0.2171
    123      2Q4    0.1973
    124      3Q1    0.2499
    125      3Q2    0.1819
    126      3Q3    0.2613
    127      3Q4    0.2302
    128      4Q1    0.3795
    129      4Q2    0.2406
    130      4Q3    0.2486
    131      4Q4    0.2464

Using ggplot to plot it as a time series is pretty easy:

require(ggplot2)
Bu.Dis["Date"] <- NA
Bu.Dis$Date <- seq(as.Date("2001/1/1"), as.Date("2004/12/31"), by = "3 months")
ggplot(Bu.Dis, aes(Date, DIS))+geom_line()+
  theme(legend.position = "none", panel.background = element_rect(fill = "#FFFFFF", colour="#000000"), 
    panel.grid.major = element_line(colour = "grey", linetype = "dotted"))

However, when i try to use the changepoint package, I have to transform the dataframe to a timeseries vector.

Bu.Dis.ts <- ts(Bush.Dis[,2], c(2001, 1), frequency = 4)
var.Bu.Dis=cpt.var(Bu.Dis.ts, method="PELT")
plot(var.Bu.Dis)

Plotting this vector with "plot" is also pretty easy, but I want to do it with ggplot. The problem is, that var.Bu.Dis is a S4-object. How do I plot this S4 object with ggplot?

By the way, is there an easier way to transform my entries in "Date" into a timeseries compatible date-format (preferable into a quarter-format like 2001-1 or 2001-Q1)?

Upvotes: 2

Views: 1504

Answers (2)

Adam Erickson
Adam Erickson

Reputation: 6363

I would look to ggfortify, which can parse ts objects into ggplot2: http://rpubs.com/sinhrks/plot_ts

Another nice thing about ggfortify is that it support timeseries statistics: http://rpubs.com/sinhrks/plot_tsstats

To format the date field as year and quarter, try the following:

library(zoo)
yq <- as.yearqtr(Bu.Dis$Date, format = "%Y/%m/%d")
format(yq, format = "%Y-0%q")

In combination, this should give you what you need. Assuming you have multiple observations per quarter, you may need to do some aggregation with stat_summary().

Cheers,

Adam

Upvotes: 1

feder80
feder80

Reputation: 1351

I think I answered this question on my own (at least most the problems I had): ggplot cannot plot ts data. So you have to transform it.

ggplot(melt(data.frame(time=as.numeric(time([email protected])), [email protected]), id.vars="time"), aes(time, value)) + 
  geom_line() +
  geom_vline(xintercept = cpts.ts(var.Bu.Dis), colour="red", linetype = "dotdash")

geom_vline is used to plot the "change point".

But I still get this message: "Don't know how to automatically pick scale for object of type ts. Defaulting to continuous"

But there is still one problem left: In the x-axis, only the years are shown, but not the quarters. How can I change that?

Upvotes: 1

Related Questions