R For Data Science Cheat Sheet: xts

eXtensible Time Series (xts) is a powerful package that provides an extensible time series class, enabling uniform handling of many R time series classes by extending zoo.

Load the package as follows:

library(xts)

Xts Objects

xts objects have three main components:

• coredata: always a matrix for xts objects
• index: vector of any Date, POSIXct, chron, yearmon, yearqtr, or DateTime classes
• xtsAttributes: arbitrary attributes

Creating xts Objects

xts1 <- xts(x=1:10, order.by=Sys.Date()-1:10)
data <- rnorm(5)
dates <- seq(as.Date("2017-05-01"), length=5, by="days")
xts2 <- xts(x=data, order.by=dates)
xts3 <- xts(x=rnorm(10), order.by=as.POSIXct(Sys.Date()+1:10), born=as.POSIXct("1899-05-08")
xts4 <- xts(x=1:10, order.by=Sys.Date()+1:10)

Convert To And From xts

data(AirPassengers)
xts5 <- as.xts(AirPassengers)

Import From Files

dat <- read.csv(tmp_file)
xts(dat, order.by=as.Date(rownames(dat),"%m/%d/%Y"))
dat_zoo <- read.zoo(tmp_file, index.column=0, sep=",", format="%m/%d/%Y")
dat_zoo <- read.zoo(tmp,sep=",",FUN=as.yearmon)
dat_xts <- as.xts(dat_zoo)

Inspect Your Data

Extract core data of objects

core_data <- coredata(xts2)

Extract index from objects

index(xts1)

Class Attributes

Get index class

indexClass(xts2)

Replacing index class

indexClass(convertIndex(xts,'POSIXct'))

Get index class

indexTZ(xts5)

Change format of time display

indexFormat(xts5) <- "%Y-%m-%d"

Time Zones

Change the time zone

tzone(xts1) <- "Asia/Hong_Kong"

Extract the current time zone

tzone(xts1)

Periods, Periodicity & Timestamps

Estimate frequency of observations

periodicity(xts5)

Convert xts5 to yearly OHLC

to.yearly(xts5)

Convert xts3 to monthly OHLC

to.monthly(xts3)

Convert xts5 to quarterly OHLC

to.quarterly(xts5)

Convert to quarterly OHLC

to.period(xts5,period="quarters")

Convert to yearly OHLC

to.period(xts5,period="years")

Count the months in xts5

nmonths(xts5)

Count the quarters in xts5

nquarters(xts5)

Count the years in xts5

nyears(xts5)

Make index unique

make.index.unique(xts3,eps=1e-4)

Remove duplicate times

make.index.unique(xts3,drop=TRUE)

Round index time to the next n seconds

align.time(xts3, n=3600)

Other Useful Functions

Extract raw numeric index of xts1

.index(xts4)

value of weekday in index of xts3

.indexwday(xts3)

Value of hour in index of xts3

.indexhour(xts3)

Extract first observation of xts3

start(xts3)

Extract last observation of xts4

end(xts4)

Display structure of xts3

str(xts3) 

Extract raw numeric index of xts1

time(xts1)

First part of xts2

head(xts2)

Last part of xts2

tail(xts2)

Export xts Objects

data_xts <- as.xts(matrix)
tmp <- tempfile()
write.zoo(data_xts,sep=",",file=tmp)

Replace & Update

Replace values in xts2 on dates with 0

xts2[dates] <- 0 

Replace dates from 1961 with NA

xts5["1961"] <- NA

Replace the value at 1 specific index with NA

xts2["2016-05-02"] <- NA 

Applying Functions

Locate endpoints by time 

ep1 <- endpoints(xts4,on="weeks",k=2)  
ep2 <- endpoints(xts5,on="years") 

Calculate the yearly mean 

period.apply(xts5,INDEX=ep2,FUN=mean)

Split xts5 by year

xts5_yearly <- split(xts5,f="years")

Create a list of yearly means   

lapply(xts5_yearly,FUN=mean) 

Find the last observation in each year in xts5

do.call(rbind,  lapply(split(xts5,"years"), function(w) last(w,n="1 month"))) 

Calculate cumulative annual passengers

do.call(rbind,  lapply(split(xts5,"years"),  cumsum)) 

Apply standard deviation to rolling margins of xts5

rollapply(xts5, 3, sd)  

Selecting, Subsetting & Indexing

Select

mar55 <- xts5["1955-03"] #Get value for March 1955

Subset

Get all data from 1954 

xts5_1954 <- xts5["1954"] 

Extract data from Jan to March ‘54 

xts5_janmarch <- xts5["1954/1954-03"]

Get all data until March ‘54  

xts5_janmarch <- xts5["/1954-03"] 

Subset xts4 using ep2

xts4[ep1] 

first() and last()

first(xts4,'1 week') #Extract first 1 week
first(last(xts4,'1 week'),'3 days') #Get first 3 days of the last week of data

Indexing

Extract rows with the index of xts3

xts2[index(xts3)] 

Extract rows using the vector days

days <- c("2017-05-03","2017-05-23") 
xts3[days] 

Extract rows using days as POSIXct

xts2[as.POSIXct(days,tz="UTC")] 

Index of weekend days

index<-which(.indexwday(xts1)==0|.indexwday(xts1)==6) 

Extract weekend days of xts1

xts1[index] 

Missing Values

Omit NA values in xts5

na.omit(xts5) 

Fill missing values in xts2 using last observation

xts_last <- na.locf(xts2) 

Fill missing values in xts2 using next observation

xts_last <- na.locf(xts2,  fromLast=TRUE)

Interpolate NAs

na.approx(xts2)  

Arithmetic Operations

coredata() or as.numeric()

xts3 + as.numeric(xts2) #Addition
xts3 * as.numeric(xts4) #Multiplication
coredata(xts4) - xts3 #Subtraction
coredata(xts4) / xts3 #Division

Shifting Index Values

Period-over-period differences 

xts5 - lag(xts5) 

Lagged differences

diff(xts5,lag=12,differences=1) 

Reindexing

xts1 + merge(xts2,index(xts1),fill=0) #Addition
xts1 - merge(xts2,index(xts1),fill=na.locf) #Subtraction

Merging

merge(xts2,xts1,join='inner') # Inner join of xts2 and xts1
merge(xts2,xts1,join='left',fill=0) #Left join of xts2 and xts1,
rbind(xts1, xts4)

Going Further

Want to know more about xts? Check out DataCamp's Manipulating Time Series Data in R with xts & zoo course!