New variable in dataframe

Question

I have a data frame of time-series prices and I want a new column within the data frame which is the average of the previous 10 observations of the prices. Therefore this new variable must start in row 10 and be rolling, i.e. observation 10 is the average of the previous 10 prices (observations 1-10), observation 11 is the average of the previous 10 prices (observations 2-11) and so on.

Answer 1

Use running() from gtools package:

#dummy data
set.seed(123)
df <- data.frame(x=round(runif(12,1,10)))

require(gtools)
#get running mean starting on 10th row
df$RunningMean <- running(df$x,width = 10,pad=TRUE)
df

#output
    x RunningMean
1   4          NA
2   8          NA
3   5          NA
4   9          NA
5   9          NA
6   1          NA
7   6          NA
8   9          NA
9   6          NA
10  5         6.2
11 10         6.8
12  5         6.5

Or if you prefer to use gtools and dplyr together. The same result.

library(gtools); library(dplyr)
df <- data.frame(x=round(runif(12,1,10)))
df %>% mutate(RunningMean = running(x, width=10, pad=TRUE, fun=mean))

Answer 2

Or simply use filter which allows you to do that without loading extra packages :

set.seed(123)
df <- data.frame(x=round(runif(12,1,10)))
n <- 10

df <- within(df,{
  runmean <- filter(x,rep(1/n,n), sides=1)
})

Another tip: if you want to do that within a data frame and you don't want to be typing those dollar signs the whole time, use the convenient function within() as I did in the example. Don't forget to put the argument sides=1, otherwise you'll take the mean centered around the value you're at (which is the default).

Answer 3

You could use zoo (Using @KFB's example). Using the align argument, you can specify left, right or center for the NAs to occupy.

 library(zoo) 
 rollmean(df$x, k=10,na.pad=TRUE, align="right")
 #[1]  NA  NA  NA  NA  NA  NA  NA  NA  NA 6.2 6.8 6.5

Answer 4

#the first 9 rows are blank
for (i in 1:9)
{
     average[i]=NA
}
for (i in 1:(length(prices)-9))
{
     average[i+9]=mean(prices[i:(i+9)])
}