conditionally remove a column in a data frame?-CodePudding

I would like to remove the column in which there is an "o"

  Var1 Var2 Var3
1    o    o    o
2 Var1    o    o
3    o Var2    o
4 Var1 Var2    o
5    o    o    o 
6 Var1    o    o
7    o Var2    o
8 Var1 Var2    o

The problem is that I have several tables and the position of the column that contains the "o" is different in each data frame. I would like to know a way to conditionally remove the columns with "o", since in the different tables the column is in a different position. Could be by evaluating the first row since the whole column is full of "o".

The output would look like

  Var1 Var2 
1    o    o    
2 Var1    o    
3    o Var2    
4 Var1 Var2    
5    o    o     
6 Var1    o    
7    o Var2    
8 Var1 Var2

CodePudding user response：

Do you need

df[,colMeans(df=="o")<1]

that is remove all columns which have all 'o's?

CodePudding user response：

Here are couple of options -

Base R Filter.

res <- Filter(function(x) any(x != 'o'), df)
res

#  Var1 Var2
#1    o    o
#2 Var1    o
#3    o Var2
#4 Var1 Var2
#5    o    o
#6 Var1    o
#7    o Var2
#8 Var1 Var2

Using purrrs keep and discard -

purrr::keep(df, ~any(.x != 'o'))
purrr::discard(df, ~all(.x == 'o'))