How to extract a count of weekdays from a vector in R

Question

I have a set of data from historical stock prices for publicly traded companies on the NASDAQ and NYSE. I've used the following code to download data on Amazon's stock price and get the times the data was sampled.

library(quantmod)
amzn <- getSymbols("AMZN",auto.assign=FALSE)
sampleTimes <- index(amzn)

Then I used grepl to get a count of how many values were collected in 2012. Now I need to get a count of how many of those values occurred on Mondays.

I've tried several different approaches, but none has been effective. I'm thinking that the lubridate package should be useful, but I can't figure out just how. POSIXlt should also be helpful, but I also can't think of an effective way to use it.

This must be a very simple problem, but I can't seem to wrap my brain around it.

Please help.

Try to convert date to days using `weekdays()` and then tabulate your column... — vrajs5, May 28 '14 at 05:55
This is a quiz question from Coursera's Data Science Course. — Brett DeWoody, Apr 09 '15 at 02:16
In 2016 this is still a quiz question from one of Coursera's Data Science courses. — Rick Henderson, Mar 24 '16 at 02:04

Tyler Rinker · Answer 1 · 2014-05-28T20:28:23.467

3

I think this is what you're after:

table(weekdays(sampleTimes))

##    Friday    Monday  Thursday   Tuesday Wednesday 
##       373       350       376       381       382

Edit per the OP's comments:

do.call(rbind, lapply(split(sampleTimes, substring(sampleTimes, 1, 4)), function(x) {
    table(weekdays(x))
}))

##      Friday Monday Thursday Tuesday Wednesday
## 2007     51     48       51      50        51
## 2008     50     48       50      52        53
## 2009     49     48       51      52        52
## 2010     50     47       51      52        52
## 2011     51     46       51      52        52
## 2012     51     47       51      50        51
## 2013     51     48       50      52        51
## 2014     20     18       21      21        20

edited May 28 '14 at 20:28

answered May 28 '14 at 05:57

Tyler Rinker

108,132
65
322
519

This is perfect, except that I'm trying to get a list of instances within sampleTimes that occurred on Mondays in the year 2012. So there seems to be one missing piece. – user3682294 May 28 '14 at 18:17
I thought you said you could do the years? See my edit. – Tyler Rinker May 28 '14 at 20:25
Yes, sorry. I was able to extract the years, but couldn't figure out how to write a function that would BOTH extract the year 2012 and instances of Mondays. – user3682294 May 28 '14 at 20:43

Rahul Premraj · Answer 2 · 2014-05-28T20:38:50.787

1

You can quite simply achieve this using the lubridate package as below:

Edit: replaced nrow() to length()

> library(lubridate)
> sampleTimes <- ymd(sampleTimes)
> length(subset(sampleTimes, year(sampleTimes) == 2012 & wday(sampleTimes, label=T) == "Mon"))
[1] 47

edited May 28 '14 at 20:38

answered May 28 '14 at 06:00

Rahul Premraj

1,595
14
13

Hi Rahul,This only seems to return 'NULL'. I believe it should return a count. – user3682294 May 28 '14 at 18:18
Sorry, my bad. Should have used length(). – Rahul Premraj May 28 '14 at 20:39

score 1 · Answer 3 · answered May 28 '14 at 20:45

1

Just use weekdays(.) == "Monday" as a logical index into sampleTimes and add the requirement that they be in 2012:

 head( sampleTimes[ weekdays(sampleTimes)=="Monday"] )
[1] "2007-01-08" "2007-01-22" "2007-01-29" "2007-02-05" "2007-02-12" "2007-02-26"

> head( sampleTimes[ weekdays(sampleTimes)=="Monday" & substr(sampleTimes, 1,4)=="2012"] )
[1] "2012-01-09" "2012-01-23" "2012-01-30" "2012-02-06" "2012-02-13" "2012-02-27"

answered May 28 '14 at 20:45

IRTFM

258,963
21
364
487

What do I do to get a count, rather than a list of occurrences? – user3682294 May 28 '14 at 20:50
`length` .... Or: `sum( weekdays(sampleTimes)=="Monday" & substr(sampleTimes, 1,4)=="2012" )` – IRTFM May 28 '14 at 20:55

How to extract a count of weekdays from a vector in R

3 Answers3