从两个数据帧中减去每两列的平均值R

Question

Suppose I have two data frames as follows:

df1 <- data.frame(ceiling(runif(10,1,10)), ceiling(runif(10,1,10)), ceiling(runif(10,1,10)))
colnames(df1) <- c("V1","V2","V3")
df2 <- data.frame(ceiling(runif(10,1,10)), ceiling(runif(10,1,10)), ceiling(runif(10,1,10)))
colnames(df2) <- c("V1","V2","V3")

Using this dummy data, I want to create a new dataframe with 1 column and 3 rows:

         V1    

1  mean(df1$V1) - mean(df2$V1)
2  mean(df1$V2) - mean(df2$V2)
3  mean(df1$V3) - mean(df2$V3)

I also want to create another dataframe as follows:

         V1    

1  wilcox.test(df1$V1,df2$V1)$p.value
2  wilcox.test(df1$V2,df2$V2)$p.value
3  wilcox.test(df1$V3,df2$V3)$p.value

My real data has 54 columns, so for my data each dataframe would be of 54 rows.

Answer 1

Means:

data.frame(mean = colMeans(df1) - colMeans(df2))
#    mean
# V1  1.4
# V2  2.0
# V3  1.4

P-values:

data.frame(
    p.value = mapply(function(x, y) wilcox.test(x, y)$p.value, df1, df2)
)
#       p.value
# V1 0.32060365
# V2 0.07784363
# V3 0.21779915

Answer 2

Q1

data.frame(mean=sapply(df1, mean)-sapply(df2,mean))

Q2

out <- NULL
for(i in 1:ncol(df1)) out[[i]] <- wilcox.test(df1[,i], df2[,i])$p.value
data.frame(p=unlist(out))

Answer 3

You can do it using a vector of ones:

 m1 =  (t(df1) %*% rep(1, nrow(df1))) / nrow(df1) # Equivalent to a mean
 m2 =  (t(df2) %*% rep(1, nrow(df2))) / nrow(df2) 

m1-m2

Answer 4

Here's a tidyverse approach to create a table with info about the tests you've performed:

# for reproducibility
set.seed(215)

# example datasets
df1 <- data.frame(ceiling(runif(10,1,10)), ceiling(runif(10,1,10)), ceiling(runif(10,1,10)))
colnames(df1) <- c("V1","V2","V3")
df2 <- data.frame(ceiling(runif(10,1,10)), ceiling(runif(10,1,10)), ceiling(runif(10,1,10)))
colnames(df2) <- c("V1","V2","V3")

library(tidyverse)

list(df1, df2) %>%                     # put your dataframes in a list
  map_df(data.frame, .id = "df") %>%   # create a dataframe with an id value for each dataset
  tbl_df() %>%                         # for visualisation purposes only
  gather(v, x, -df) %>%                # reshape data
  nest(-v) %>%                         # nest data
  mutate(w.t = map(data, ~wilcox.test(.x$x ~ .x$df)),    # perfom wilcoxon test
         pval = map_dbl(w.t, "p.value"),                 # extract p value
         mean_diff = map_dbl(data, ~mean(.x$x[.x$df==1])-mean(.x$x[.x$df==2]))) # calculate mean difference

# # A tibble: 3 x 5
#   v     data              w.t           pval mean_diff
#   <chr> <list>            <list>       <dbl>     <dbl>
# 1 V1    <tibble [20 x 2]> <S3: htest> 0.730      0.600
# 2 V2    <tibble [20 x 2]> <S3: htest> 0.145     -1.8  
# 3 V3    <tibble [20 x 2]> <S3: htest> 0.0295     2.8 

Column v represents your variables (initial columns).

Column data includes the variables used for the corresponding test.

Column wt includes the test output.

Column pval is the extracted p value from each test.

Column mean_diff is the mean difference.

If you save the above process as results you'll be able to use results$wt and see the test outputs