Writing loop or function for various t-tests

Question

all.

I am very new to writing loops or functions in R, and I still haven't really understood how to do that. Currently, I need to write a loop/function (not sure which one would be better) to perform several t-tests with different data frames.

I have data that is similar to this:

set.seed(694)
df_1_08 <- data.frame(
  year = 2008,
  a = runif(100, 0, 100),
  b = runif(100, 0, 100),
  c = runif(100, 0, 100),
  d = runif(100, 0, 100)
)

df_1_09 <- data.frame(
  year = 2009,
  a = rnorm(100, 0, 1),
  b = rnorm(100, 0, 1),
  c = rnorm(100, 0, 1),
  d = rnorm(100, 0, 1)
)


df_1_10 <- data.frame(
  year = 2010,
  a = rnorm(100, 0, 1),
  b = rnorm(100, 0, 1),
  c = rnorm(100, 0, 1),
  d = rnorm(100, 0, 1)
)


df_2_08 <- data.frame(
  year = 2008,
  a = rnorm(100, 0, 1),
  b = rnorm(100, 0, 1),
  c = rnorm(100, 0, 1),
  d = rnorm(100, 0, 1)
)


df_2_09 <- data.frame(
  year = 2009,
  a = rnorm(100, 0, 1),
  b = rnorm(100, 0, 1),
  c = rnorm(100, 0, 1),
  d = rnorm(100, 0, 1)
)


df_2_10 <- data.frame(
  year = 2010,
  a = rnorm(100, 0, 1),
  b = rnorm(100, 0, 1),
  c = rnorm(100, 0, 1),
  d = rnorm(100, 0, 1)
)

# Write Loop to do t-test between dfs 08, 09, 10 comparing columns a, b, c, d and storing the full results in a df

Essentially, what I need to do with this data is to run t-tests for specific columns each year (2008, 2009, 2010) so that df_1_08 runs a t-test with df_2_08 in all columns (a, b, c, d) and then store these in a data frame (with the t-statistic, p-value, etc stored in it). This sounds like a job perfect for a loop. But I also need to do this for each of the years (2008, 2009, and 2010) and store the results in separate data frames, so this sounds like a job perfect for a function.

I'm unsure about how to write either, so I figured I'd ask for some help in writing these loops/functions. Thanks in advance for any help you might provide.

I could also have the dataframes combined into one large df with a column identifying the original data frame number (i.e. df1 or df2) and one column identifying the data frame year (i.e. 2008, 2009, 2010). It would look like this:

df1 <- rbind(df_1_08, df_1_09, df_1_10)
df1$ID <-1
df2 <- rbind(df_1_08, df_1_09, df_1_10)
df2$ID <- 2

master.df <- rbind(df1, df2)

I'm not sure if it would be easier to write a loop/function to run the t.tests with the master.df. In that df, I would essentially need to do the following within a loop or function:

1. Subset master.df into df1 and df2
2. Subset df1 and df2 in years
3. Run t.test for columns a, b, c, and d for each year
4. Store all relevant t.test outputs (i.e. t-statistic, p-value, etc) in a data.frame that I can then print.

Answer 1

How about:

df_1_08 <- data.frame(year = 2008, a = runif(100, 0, 100), b = runif(100, 0, 100), c = runif(100, 0, 100), d = runif(100, 0, 100))
df_1_09 <- data.frame(year = 2009, a = runif(100, 0, 100), b = runif(100, 0, 100), c = runif(100, 0, 100), d = runif(100, 0, 100))
df_1_10 <- data.frame(year = 2010, a = runif(100, 0, 100), b = runif(100, 0, 100), c = runif(100, 0, 100), d = runif(100, 0, 100))
df_2_08 <- data.frame(year = 2008, a = runif(100, 0, 100), b = runif(100, 0, 100), c = runif(100, 0, 100), d = runif(100, 0, 100))
df_2_09 <- data.frame(year = 2009, a = runif(100, 0, 100), b = runif(100, 0, 100), c = runif(100, 0, 100), d = runif(100, 0, 100))
df_2_10 <- data.frame(year = 2010, a = runif(100, 0, 100), b = runif(100, 0, 100), c = runif(100, 0, 100), d = runif(100, 0, 100))

dfs_1.names <- ls()[grep("df_1", ls())]
dfs_2.names <- ls()[grep("df_2", ls())]
dfs_1.list <-lapply(dfs_1.names, get)
dfs_2.list <- lapply(dfs_2.names, get)

#in case you want to try the matrix
dfs_1.mtrx <- do.call("rbind",dfs_1.list)
dfs_2.mtrx <- do.call("rbind",dfs_2.list)

years <- intersect(unique(dfs_1.mtrx[,"year"]),unique(dfs_2.mtrx[,"year"]))
# [1] 2008 2009 2010
columns <- intersect(colnames(dfs_1.mtrx[,-1]),colnames(dfs_2.mtrx[,-1]))
# [1] "a" "b" "c" "d"

df.ttest <- as.data.frame(matrix(NA, ncol = 8, nrow = length(years)*length(columns)))
colnames(df.ttest) <- c("year","column","tstat","p.value","degreesf","low.conf","up.conf","data.name")
count = 0
for(i in 1:length(years)){
  for(j in columns){
    ttest <- t.test(dfs_1.list[[i]][j], dfs_2.list[[i]][j])
    ttest$data.name <- paste(paste0("df_1_",years[i]-2000,"$",j),"and",
                             paste0("df_2_",years[i]-2000,"$",j))
    count <- count + 1
    df.ttest[count, "year"]     <- years[i]
    df.ttest[count, "column"]   <- j
    df.ttest[count, "tstat"]    <- ttest$statistic
    df.ttest[count, "p.value"]  <- ttest$p.value
    df.ttest[count, "degreesf"] <- ttest$parameter
    df.ttest[count, "low.conf"] <- ttest$conf.int[1]
    df.ttest[count, "up.conf"]  <- ttest$conf.int[2]   
    df.ttest[count, "data.name"] <- ttest$data.name
  }
}
df.ttest

Which looks like:

   year column      tstat    p.value degreesf    low.conf   up.conf               data.name
1  2008      a  1.0607688 0.29008725 197.9914  -3.7038792 12.327117   df_1_8$a and df_2_8$a
2  2008      b  0.3311722 0.74086573 197.3689  -6.6956039  9.398291   df_1_8$b and df_2_8$b
3  2008      c  1.0410813 0.29910773 197.9405  -3.7582835 12.164152   df_1_8$c and df_2_8$c
4  2008      d  1.2623350 0.20834791 193.4532  -2.9384999 13.387911   df_1_8$d and df_2_8$d
5  2009      a -0.5764091 0.56500626 194.1686 -10.1442158  5.555762   df_1_9$a and df_2_9$a
6  2009      b -1.5222524 0.12954190 197.9248 -14.4317793  1.857603   df_1_9$b and df_2_9$b
7  2009      c -0.1744245 0.86171283 195.0217  -8.6590932  7.251902   df_1_9$c and df_2_9$c
8  2009      d  0.0839337 0.93319409 197.6654  -7.5768817  8.250526   df_1_9$d and df_2_9$d
9  2010      a  1.9125742 0.05724768 197.7406  -0.2353887 15.378495 df_1_10$a and df_2_10$a
10 2010      b  0.9024489 0.36792603 196.0224  -4.0977460 11.011904 df_1_10$b and df_2_10$b
11 2010      c -0.9735756 0.33145768 197.5899 -12.2641333  4.157135 df_1_10$c and df_2_10$c
12 2010      d  0.8721498 0.38418378 197.8601  -4.5311820 11.717207 df_1_10$d and df_2_10$d