[R] Searching for specific values in a matrix

[Steve Lianoglou]

On Jul 21, 2009, at 3:27 PM, Mehdi Khan wrote:

> I understand your explanation about the test for even numbers.   
> However I am still a bit confused as to how to go about finding a  
> particular value.  Here is an example data set
>
> col #          attr1    attr2   attr 3    LON        LAT
> 17209         D        NA    NA -122.9409 38.27645
> 17210        BC        NA    NA -122.9581 38.36304
> 17211         B        NA    NA -123.6851 41.67121
> 17212        BC        NA    NA -123.0724 38.93073
> 17213         C        NA    NA -123.7240 41.84403
> 17214      <NA>       464    NA -122.9430 38.30988
> 17215         C        NA    NA -123.4442 40.65369
> 17216        BC        NA    NA -122.9389 38.31551
> 17217         C        NA    NA -123.0747 38.97998
> 17218         C        NA    NA -123.6580 41.59610
> 17219         C        NA    NA -123.4513 40.70992
> 17220         C        NA    NA -123.0901 39.06473
> 17221        BC        NA    NA -123.0653 38.94845
> 17222        BC        NA    NA -122.9464 38.36808
> 17223      <NA>       464    NA -123.0143 38.70205
> 17224      <NA>        NA     5 -122.8609 37.94137
> 17225      <NA>        NA     5 -122.8628 37.95057
> 17226      <NA>        NA     7 -122.8646 37.95978

For future reference, perhaps paste this in a way that's easy for us  
to paste into a running R session so we can use it, like so:

df <- data.frame(
coln=c(17209, 17210, 17211, 17212, 17213, 17214, 17215, 17216, 17217,  
17218, 17219, 17220, 17221, 17222, 17223, 17224, 17225, 17226),
attr1 
= 
c 
("D 
","BC","B","BC","C",NA,"C","BC","C","C","C","C","BC","BC",NA,NA,NA,NA),
attr2=c( NA,NA,NA,NA,NA,464,NA,NA,NA,NA,NA,NA,NA,NA,464,NA,NA,NA),
attr3=c(NA,NA,NA,NA,NA,NA,NA,NA,NA,NA,NA,NA,NA,NA,NA,5,5,7),
LON 
= 
c 
( -122.9409 
,-122.9581 
,-123.6851 
,-123.0724 
,-123.7240 
,-122.9430 
,-123.4442 
,-122.9389 
,-123.0747 
,-123.6580 
,-123.4513 
,-123.0901,-123.0653,-122.9464,-123.0143,-122.8609,-122.8628,-122.8646),
LAT 
= 
c 
(38.27645,38.36304,41.67121,38.93073,41.84403,38.30988,40.65369,38.31551,38.97998,41.59610,40.70992,39.06473,38.94845,38.36808,38.70205,37.94137,37.95057,37.95978 
))

> If I wanted to find the row with Lat = 37.95978

Using an "indexing vector":

R> lats <- df$LAT == 37.95978
# or with the %~% from before:
# lats <- df$LAT %~% 37.95978
R> df[lats,]
     coln attr1 attr2 attr3       LON      LAT
18 17226  <NA>    NA     7 -122.8646 37.95978

Using the "subset" function:

R> subset(df, LAT == 37.95978)
     coln attr1 attr2 attr3       LON      LAT
18 17226  <NA>    NA     7 -122.8646 37.95978

> , how would i do that?  How would  I find the rows with BC?

R> subset(df, attr1 == 'BC')
     coln attr1 attr2 attr3       LON      LAT
2  17210    BC    NA    NA -122.9581 38.36304
4  17212    BC    NA    NA -123.0724 38.93073
8  17216    BC    NA    NA -122.9389 38.31551
13 17221    BC    NA    NA -123.0653 38.94845
14 17222    BC    NA    NA -122.9464 38.36808


If you try with an "indexing vector" the NA's will trip you up:

R> df[df$attr1 == 'BC',]
       coln attr1 attr2 attr3       LON      LAT
2    17210    BC    NA    NA -122.9581 38.36304
4    17212    BC    NA    NA -123.0724 38.93073
NA      NA  <NA>    NA    NA        NA       NA
8    17216    BC    NA    NA -122.9389 38.31551
13   17221    BC    NA    NA -123.0653 38.94845
14   17222    BC    NA    NA -122.9464 38.36808
NA.1    NA  <NA>    NA    NA        NA       NA
NA.2    NA  <NA>    NA    NA        NA       NA
NA.3    NA  <NA>    NA    NA        NA       NA
NA.4    NA  <NA>    NA    NA        NA       NA

So you could do something like:

 > df[df$attr1 == 'BC' & !is.na(df$attr1),]
     coln attr1 attr2 attr3       LON      LAT
2  17210    BC    NA    NA -122.9581 38.36304
4  17212    BC    NA    NA -123.0724 38.93073
8  17216    BC    NA    NA -122.9389 38.31551
13 17221    BC    NA    NA -123.0653 38.94845
14 17222    BC    NA    NA -122.9464 38.36808

HTH,
-steve

--
Steve Lianoglou
Graduate Student: Physiology, Biophysics and Systems Biology
Weill Medical College of Cornell University

Contact Info: http://cbio.mskcc.org/~lianos/contact