Please I need to calculate the SD by several variable rows (columns),

Question

I tried to loop, apply, as I am learning, I did not get the solution, the idea and creates a function that calculates the sd this function provides the calculation per line:

f_sd_subgrupo <- function(dados){
  sd_subgrupo <- dados %>% 
    select(starts_with("amostra"))%>%
    apply(1, sd, na.rm = TRUE)
}

What I seek is this:

amostra <- data.frame(
                      subgrupo = c(rep(1:4)),
                      cartaid = 1,
                      amostra.1 = c (50.,49,48,47),
                      amostra.2 = c(51.,49.,48,49),
                      amostra.3 = c(50.,51.,48.,49),
                      amostra.4 = c(51.,50.,52.,48),
                      amostra.5 = c(49.,50.,52.,51))

Sd calculated in the described function provides the Standard Deviation to each line, the values are:

subgrupo 01 : 1.29099   SdOverall 1.29099
subgrupo 02 : 1.50000   sdoverall 1.457738
subgrupo 03 : 1.29099   sdoverall 1.356801
subgrupo 04 : 1.707825  sdoverall 1.460593

I am looking for a function to calculate the following:

Sdoverall that accumulates to each row the values of the variables (samples) to each subgroup:

Sdoverall of subgroup 1 are all sample values (sdoverall = 1.29099).

Sdoverall of subgroup 2 are the sample values of subgroup 1 and the subgroup 2 values (1.457738).

overall of subgroup 3 are the sample values of subgroup 1, subgroup 2 and subgroup 3 (1.356801).... and continue

No and sum from subgroup 2 the values (samples) of the previous subgroup, type i in 2:n, shall be added, the number of elements thus accumulates : Subgroup 2 has the values of row 01(subgroup 1) and row 02(subgroup 2) to calculate the sample standard deviation of subgroup 2.

Thank you

---

1. Relate the item

---

Answer 1

With the question data, the following code calculates the standard deviations of

1. Line 1;
2. Lines 1 and 2;
3. Lines 1, 2 and 3;
4. Lines 1, 2, 3 and 4.

But the results do not match the results in the question.

i_col <- grep('^amostra', names(amostra))

sapply(seq_len(nrow(amostra)), function(i){
  x <- amostra[seq_len(i), i_col]
  x <- unlist(x)
  sd(x)
})
#[1] 0.8366600 0.8164966 1.3557637 1.4290225

The above code can be rewritten as a function.

1. The function has 3 arguments,
   1. the basis x, a data frame.;
   2. the columns where the standard deviation values are to be calculated;
   3. if the values NA must be removed.
2. in the first instruction, apply to each of the x an anonymous function;
3. this function extracts the i first rows and the columns in question in the object y;
4. y can be a data.frame, so it’s transformed into a vector;
5. the standard deviation is calculated and is the output of the function.

Since the base is the first argument, this function can be used in a pipe package magrittr.

sdAcum <- function(x, cols, na.rm = FALSE){
  sapply(seq_len(nrow(x)), function(i){
    y <- x[seq_len(i), cols]
    y <- unlist(y)
    sd(y, na.rm = na.rm)
  })
}

sdAcum(amostra, i_col)    # mesmo resultado

With the package dplyr, the first way to call the function returns a vector, the second way a data frame..

amostra %>% sdAcum(i_col)

amostra %>%
  mutate(Sdoverall = sdAcum(., i_col))