Calculate rolling mean for each row in a data frame

Question

For a data frame that looks like:

Value  Id
 1      xxx-rrr-ttt
 78     ggg-oop-rty
 97     fad-dar-oki
 ..      ..
 ..      ..
 ..      ..

I need to calculate the rolling-mean at each row. So for a data frame s, rolling mean at each row will be represented as s.mean. Is there any inbuilt function in R that can calculate rolling-mean ? I tried the following:

rollapply(data)

but it throws an error argument "width" is missing, with no default. I could not understand this error and what is meant by the width

You can use `rollmean` from `zoo`. With `rollapply`, you need to pass the function and width argument. You can check the documentation of `?rollapply` if there is doubt. `width - numeric vector or list. In the simplest case this is an integer specifying the window width (in numbers of observations) which is aligned to the original sample according to the align argument. Alternatively, width can be a list regarded as offsets compared to the current time, see below for details.` — akrun, Mar 28 '18 at 02:24
@akrun I have already imported the package, but not sure how to use the function. — Jatt, Mar 28 '18 at 02:25
you probably need to pass a column name instead of complete dataframe. What is your expected output? — Ronak Shah, Mar 28 '18 at 02:30
@RonakShah How did you term it as duplicate? Any question related to `moving mean` cannot be a duplicate. — Jatt, Mar 28 '18 at 02:50
@Jatt I see no difference between the question you have asked and the one marked. If you would have read the documentation at `?rollmean` or `?rollapply` you would have got the answer yourself. Moreover, if you still think your question is different from the one marked please feel free to vote to reopen the question. — Ronak Shah, Mar 28 '18 at 02:55

score 3 · Accepted Answer · edited Jun 20 '20 at 09:12

3

We can get the same output with rollmean or rollapply

library(zoo)
rollmean(df1$Value, k = 3)
#[1]  0.13677590  0.12419375  0.22781866  0.17661827  0.51935576  0.08137071 -0.29009330 -0.43751774
rollapply(df1$Value, width = 3, FUN = mean)
#[1]  0.13677590  0.12419375  0.22781866  0.17661827  0.51935576  0.08137071 -0.29009330 -0.43751774

data

set.seed(24)
df1 <- data.frame(Value = rnorm(10))

edited Jun 20 '20 at 09:12

Community

1
1

answered Mar 28 '18 at 02:30

akrun

874,273
37
540
662

What is `3` in the rollmean and rollapply? – Jatt Mar 28 '18 at 02:32
@Jatt Updated the post – akrun Mar 28 '18 at 02:33
What does it mean by `width = 3`? I could not understand this. How did you come to 3? – Jatt Mar 28 '18 at 02:34
@Jatt It is a example to show you that for 3 values, it does the rollling mean. In your data, I am not sure what would be your rolling mean width. Are you going to take the mean for every 2 values, 3 values? – akrun Mar 28 '18 at 02:35
Okay. I get it. Is there any way I could also calculate rolling-standard deviation? – Jatt Mar 28 '18 at 02:42
@Jatt You can use `rollapply` and change `FUN` to `sd` If you need a custom function, use `RcppRoll:roll_sd` – akrun Mar 28 '18 at 02:43
1

rollapply for `sd` worked fine! – Jatt Mar 28 '18 at 02:48
Don't mean to extend this. Just wanted to ask you one last thing. To find data points that are more than (6*s.d.) away from mean, in the same data frame, does the following make sense `df[df$value > 6*df$sd + df$mean,]` ? – Jatt Mar 28 '18 at 03:17
@Jatt What is `4*` ? – akrun Mar 28 '18 at 03:19
Just removed it. It was a typo – Jatt Mar 28 '18 at 03:19
Made it `6*df$sd` – Jatt Mar 28 '18 at 03:20
I meant `df$mean - 6*df$sd` – akrun Mar 28 '18 at 03:21
Did not get the `abs` part? – Jatt Mar 28 '18 at 03:21
I meant to say about mean + or - – akrun Mar 28 '18 at 03:22
Okay. `df[df$value > (6*df$sd + df$mean) | (df$mean - 6*df$sd),]` Does it make sense? – Jatt Mar 28 '18 at 03:24
1

It should be wrapped with `abs` i.e. `abs((6*df$sd + df$mean) | (df$mean - 6*df$sd))` – akrun Mar 28 '18 at 03:25

Calculate rolling mean for each row in a data frame

1 Answers1

data