Loop with correlation

rstudio

#1

I have this matrix and following code:

# Remove NA-observations from dataset "lus" and removing row "totalsum"
lus2 <- na.omit(lus)

lus3 <- lus2[-c(10),]


# The problem now is that "laksepris" has months in the columns, while "lus" has months in the rows

laksepris2 <- laksepris %>%
  spread (Month, Pris)

test <- rbind(setDT(lus3), setDT(laksepris2), fill=TRUE)


test[10,1] <- "Pris pr.kilo"

test_round <- test %>% 
  mutate_if(is.numeric, round, digits = 2)

#-------------------------------------------


rearranget_lus <- as.data.frame(t(test_round))

rearranget_lus

# Removing first row, and renaming the columns:

lus_1 <- rearranget_lus[-c(1),]

names (lus_1) [1] <- "Finmark"
names (lus_1) [2] <- "Troms"
names (lus_1) [3] <- "Nordland"
names (lus_1) [4] <- "Nord-Trondelag"
names (lus_1) [5] <- "Sor-Trondelag"
names (lus_1) [6] <- "More og Romsdal"
names (lus_1) [7] <- "Sogn og Fjordane"
names (lus_1) [8] <- "Hordaland"
names (lus_1) [9] <- "Rogaland og Agder"
names (lus_1) [10] <- "Pris pr.kilo"

I just started using R, and I am therefore wondering how I can run a correlation between the values in "pris pr.kilo" against the values in column "Finmark". Following I would also like to loop this, so that the loop runs the correlation between "pris.pr.kilo" and all the other columns as well.
Does anyone have a suggestion to how this is done?


#2

Since I don't have access to your data, I'll create a synthetic data set.

library(tidyverse)

df <- data_frame(A = 1:12,
                 B = rev(A),
                 C = 2 * A,
                 D = 2 * B, 
                 Y = 1:12)
> df
# A tibble: 12 x 5
       A     B     C     D     Y
   <int> <int> <dbl> <dbl> <int>
 1     1    12     2    24     1
 2     2    11     4    22     2
 3     3    10     6    20     3
 4     4     9     8    18     4
 5     5     8    10    16     5
 6     6     7    12    14     6
 7     7     6    14    12     7
 8     8     5    16    10     8
 9     9     4    18     8     9
10    10     3    20     6    10
11    11     2    22     4    11
12    12     1    24     2    12

cor(A,Y) and cor(C,Y) should be 1. cor(B,Y) and cor(D,Y) should be -1.

I would break the dataframe into two pieces:

  • the portion you want to loop over (X)
  • and the portion that should stay constant (Y)
X <- select(df, -Y)
Y <- select(df, Y)

Now I can use map_df from the purrr package to feed each column of X to cor while setting the y parameter to Y. The output will be a dataframe.

library(purrr)

result <- map_df(X, cor, y = Y)
> result
# A tibble: 1 x 4
      A     B     C     D
  <dbl> <dbl> <dbl> <dbl>
1     1    -1     1    -1

#3

You can also do this with the corrr package. Using the dataset given by @Galangjs it would look like this:

library(tidyverse)
library(corrr)

df <- data_frame(A = 1:12,
                 B = rev(A),
                 C = 2 * A,
                 D = 2 * B, 
                 Y = 1:12)

#default output of correlate function
corrr_result <- correlate(df)
#> 
#> Correlation method: 'pearson'
#> Missing treated using: 'pairwise.complete.obs'

corrr_result
#> # A tibble: 5 x 6
#>   rowname     A     B     C     D     Y
#>   <chr>   <dbl> <dbl> <dbl> <dbl> <dbl>
#> 1 A          NA    -1     1    -1     1
#> 2 B          -1    NA    -1     1    -1
#> 3 C           1    -1    NA    -1     1
#> 4 D          -1     1    -1    NA    -1
#> 5 Y           1    -1     1    -1    NA

# look only at the desired comparison
corrr_result %>% 
  filter(rowname == "Y") %>% 
  select(-Y)
#> # A tibble: 1 x 5
#>   rowname     A     B     C     D
#>   <chr>   <dbl> <dbl> <dbl> <dbl>
#> 1 Y           1    -1     1    -1

Created on 2018-08-29 by the reprex package (v0.2.0).