R help - Jun 2014 - selecting columns based on partial names

Hello,

I have a data frame with > 5000 columns and I'd like to be able to make 
subsets of that data frame made up of certain columns by using part of 
the column names. I've had a surprisingly hard time finding something 
that works by searching online.

For example, lets say I have a data frame (df) of 2 obs. of 6 variables. 
The 6 variables are called "1940_tmax", "1940_ppt",
"1940_tmin",
"1941_tmax", "1941_ppt", "1941_tmin". I want to
create a new data frame
with only the variables that have "ppt" in the variable (column) name,
so that it looks like this:

plot name    1940_ppt    1941_ppt
774-CL                    231               344
778-RW                  228               313

Thanks.

-- 
Christopher R. Dolanc
Post-doctoral Researcher
University of California, Davis &
University of Montana

On 19.06.2014 23:50, Chris Dolanc wrote:
> Hello,
>
> I have a data frame with > 5000 columns and I'd like to be able to
make
> subsets of that data frame made up of certain columns by using part of
> the column names. I've had a surprisingly hard time finding something
> that works by searching online.
>
> For example, lets say I have a data frame (df) of 2 obs. of 6 variables.
> The 6 variables are called "1940_tmax", "1940_ppt",
"1940_tmin",
> "1941_tmax", "1941_ppt", "1941_tmin". I want
to create a new data frame
> with only the variables that have "ppt" in the variable (column)
name,
> so that it looks like this:
>
> plot name    1940_ppt    1941_ppt
> 774-CL                    231               344
> 778-RW                  228               313
>
> Thanks.
>
df[ , grepl("_ppt$", names(df))]

Best,
Uwe Ligges

On Thu, 19 Jun 2014 02:50:20 PM Chris Dolanc wrote:
> Hello,
> 
> I have a data frame with > 5000 columns and I'd like to be able to 
make
> subsets of that data frame made up of certain columns by using 
part of
> the column names. I've had a surprisingly hard time finding 
something
> that works by searching online.
> 
> For example, lets say I have a data frame (df) of 2 obs. of 6 
variables.
> The 6 variables are called "1940_tmax", "1940_ppt",
"1940_tmin",
> "1941_tmax", "1941_ppt", "1941_tmin". I want
to create a new data
frame
> with only the variables that have "ppt" in the variable (column) 
name,
> so that it looks like this:
> 
> plot name    1940_ppt    1941_ppt
> 774-CL                    231               344
> 778-RW                  228               313
> 
Hi Chris,
One way is to get the column indices:

grep("ppt",names(df))
[1] 2 5

so,

newdf<-df[grep("ppt",names(df))]

and then you apparently want to add a column with some other 
information, so probably:

newdf<-cbind(<something_else>,
 df[grep("ppt",names(df))])

Jim