Removing bijective columns automatically from the data frame identified using whichAreBijection command from dataPreparation package using r

Question

I am having trouble doing a simple task:

Say I use these libraries and have this data frame:

library(tidyverse)
library(dataPreparation)

df <- data.frame(col1 = 1, col2 = rnorm(1e1), col3 = sample(c(1, 2), 1e1, replace = TRUE))
df$col4 <- df$col2
df$col5[df$col3 == 1] = "a"
df$col5[df$col3 == 2] = "b"
df$col6 = c("b","b","a","a","b","a","a","a","a","b")
df$col7 = "d"
df$col8 = c(3,3,5,5,3,5,5,5,5,3)
df$col9 = c("x","x","y","y","x","y","y","y","y","x")
df$col10 = c("p","p","q","p","q","q","p","p","q","q")
df$col11 = c(10.5,10.5,11.37,10.5,11.37,11.37,10.5,10.5,11.37,11.37)
df <- df %>% mutate_if(is.character,as.factor)

Using the below command,I want to remove the columns 4, 5, 7, 8, 9, 11 from df.

whichAreBijection(df)
[1] "whichAreBijection: col7 is a bijection of col1. I put it in drop list."
[1] "whichAreBijection: col4 is a bijection of col2. I put it in drop list."
[1] "whichAreBijection: col5 is a bijection of col3. I put it in drop list."
[1] "whichAreBijection: col8 is a bijection of col6. I put it in drop list."
[1] "whichAreBijection: col9 is a bijection of col6. I put it in drop list."
[1] "whichAreBijection: col11 is a bijection of col10. I put it in drop list."
[1] "whichAreBijection: it took me 0.08s to identify 6 column(s) to drop."
[1]  4  5  7  8  9 11

I can remove them manually by using

df$col4 = NULL
df$col5 = NULL
df$col7 = NULL
df$col8 = NULL
df$col9 = NULL
df$col11 = NULL

I want, however, the algorithm to automatically do it.

I tried the following first to generate the data frame m containing column numbers proposed by whichAreBijection and then eventually delete it from df, but it led me no where:

x <- whichAreBijection(df)
y <- length(x)

m <- as.data.frame(matrix(0, ncol = y, nrow = nrow(df)))
i = 1
while (i< y+1) {
  # z <- names(df)[x[i]]
  m[,i] <- df[,x[i]]
  i<- i+1
}

The above generates m with constant entries given by 4, 5, 7, 8, 9, 11

I see that using a simple command like

m[,1] <- df[,4]

perfectly replaces the first column of m with the 4th column of df.

The second trouble that I have is using the same column names in m as in df. May be this sounds a long way to do the simple task.

Why are columns not being replaced exactly in m?
How could I automatically let m choose the column names of df to be deleted as column names?
Is there a better way to avoid this mess and do a straightforward removal of column names proposed by whichAreBijection?

score 0 · Answer 1 · answered Apr 10 '20 at 17:35

I was able to tackle question 1 using the following:

x <- whichAreBijection(df)
y <- length(x)
m <- as.data.frame(matrix(0, ncol = y, nrow = nrow(df)))
i = 1
while (i< y+1) {
    m[,i] <- df[,x[i], with = FALSE]
    i<- i+1
}

It seems the named column index value does not work, whereas the integer column index works on the right hand side, I mean x[i]. This trouble can be avoided by setting with = FALSE at the end.

Question 2 is the next challenge.

Removing bijective columns automatically from the data frame identified using whichAreBijection command from dataPreparation package using r

1 Answers1