Calculating Summary Statistics by Day/Week/Month

Question

In R, I have a data-frame of various statistics recorded throughout the day. (For example, heart-rate) . The time-stamps for each measurement-entry are automatically created, and I have already converted them into a POSIXt class element.

The number of observations varies from day to day.

I am wondering how I can calculate summary statistics by day/week/month.

Answer 1

Use tapply and format.

eg.

> tst<-data.frame(date=as.POSIXct(runif(1000)*31557600,origin="2010/8/9"),value=runif(1000))

> tapply(tst$value,format(tst$date,"%a"),summary)
$Fri
    Min.  1st Qu.   Median     Mean  3rd Qu.     Max. 
0.001545 0.238900 0.499600 0.484700 0.697000 0.996400 

$Mon
   Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
0.02029 0.25100 0.49100 0.49910 0.75530 0.99120 

$Sat
    Min.  1st Qu.   Median     Mean  3rd Qu.     Max. 
0.003557 0.245600 0.493600 0.499200 0.754600 0.996200 

$Sun
   Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
0.01867 0.22340 0.52750 0.51260 0.80500 0.97760 

$Thu
    Min.  1st Qu.   Median     Mean  3rd Qu.     Max. 
0.003691 0.281200 0.600600 0.546800 0.790800 0.973000 

$Tue
    Min.  1st Qu.   Median     Mean  3rd Qu.     Max. 
0.009304 0.253400 0.488900 0.510300 0.772200 0.997100 

$Wed
    Min.  1st Qu.   Median     Mean  3rd Qu.     Max. 
0.002854 0.236200 0.400600 0.473500 0.742900 0.988600

You can replace the %a in format with other codes to suit, see ?strptime. Month is %b and weeknumber is %U.

Answer 2

You could try something like this to get summary statistics by month for the second column of your dataframe

library(plyr)
library(fBasics)
dlply(my_dataframe,.(format(date_Column, "%m %y")),function(x) basicStats(x[2]))