I have a concatenated string having column names in it:
col_names = paste(c("Age", "Sex", "City", "Name"), collapse=', ')
col_names
[1] "Age, Sex, City, Name"
I also have a data frame=df that has 100s of variables.
How can I select the columns present in my string (col_names) from the df using col_names?
We can use scan to get the vector of names, remove the trailing/leading spaces with trimws and use that to subset the dataset.
df[trimws(scan(text=col_names, sep=",", what=''))]
Or another option is str_extract
library(stringr)
df[str_extract_all(col_names, '\\w+')[[1]]]
df <- data.frame(Age= c(5, 3, 25), Sex = c("M", "F",
"M"), Value = 1:3, City= c("Brookings",
"Mumbai", "Paris"), Name = LETTERS[1:3])
Does using the strsplit function help? It will split your string into a character vector by some sort of separator. Here I use the function to subset a data frame (courtesy of akrun) by the variables in the string.
df <- data.frame(Age= c(5, 3, 25), Sex = c("M", "F",
"M"), Value = 1:3, City= c("Brookings",
"Mumbai", "Paris"), Name = LETTERS[1:3])
df[,strsplit(col_names, ", ")[[1]]]
Age Sex City Name
1 5 M Brookings A
2 3 F Mumbai B
3 25 M Paris C
If you have dataframe df and column name "Column_Name", you can access the column as a vector like this:
df[["Column_Name"]]
So, when you use the colnames(df) function, you can use the strings returned to access the columns you need. A very similar method works for multiple columns:
df[c("Column_1", "Column_2", "Column_3")]
Notice that using only one set of square brackets ([]) will return a sub-dataframe, while using two sets ([[]]) will return a vector.
In your situation, you may want to consider leaving out the paste() function. The following will work just fine:
col_names = c("Age", "Sex", "City", "Name")
col_data = df[col_names]
df[names(read.csv(text = col_names))]would work.