How to remove zero variance features in R?

Question

I have to remove variables with zero variance from this dataset which has 530 variables. I used the nearzerovar function, but it eliminates almost all the variables and leaves me with only ten variables.

 str(Dtrain)
'data.frame':   19937 obs. of  530 variables:
 $ LONGITUDE       : num  -7541 -7537 -7519 -7525 -7632 ...
 $ LATITUDE        : num  4864921 4864934 4864950 4864934 4864982 ...
 $ FLOOR           : Factor w/ 5 levels "0","1","2","3",..: 3 3 3 3 1 3 3 3 3 3 ...
 $ BUILDINGID      : Factor w/ 3 levels "0","1","2": 2 2 2 2 1 2 2 2 2 2 ...
 $ SPACEID         : Factor w/ 123 levels "1","2","3","4",..: 31 31 28 27 47 30 28 26 ...
 $ RELATIVEPOSITION: Factor w/ 2 levels "1","2": 2 2 2 2 2 2 2 2 2 1 ...
 $ USERID          : int  2 2 2 2 11 2 2 2 2 2 ...
 $ PHONEID         : int  23 23 23 23 13 23 23 23 23 23 ...
 $ TIMESTAMP       : int  1371713733 1371713691 1371714095 1371713807 1369909710 ...
 $ IDidentifier    : Factor w/ 905 levels "0_0_102_2","0_0_106_2",..: 400 400 394 392 16 ...
 $ WAP001          : int  100 100 100 100 100 100 100 100 100 100 ...
 $ WAP002          : int  100 100 100 100 100 100 100 100 100 100 ...
 $ WAP003          : int  100 100 100 100 100 100 100 100 100 100 ...
 $ WAP004          : int  100 100 100 100 100 100 100 100 100 100 ...
 $ WAP005          : int  100 100 100 100 100 100 100 100 100 100 ...
 $ WAP006          : int  100 100 100 100 100 100 100 100 100 100 ...
 $ WAP007          : int  100 100 100 100 100 100 100 100 100 100 ...
 $ WAP008          : int  100 100 -97 100 100 100 100 100 100 100 ...
 $ WAP009          : int  100 100 100 100 100 100 100 100 100 100 ...
 $ WAP010          : int  100 100 100 100 100 100 100 100 100 100 ...
 $ WAP011          : int  100 100 100 100 100 100 100 100 100 100 ...
 $ WAP012          : int  100 100 100 100 100 100 100 100 100 100 ...
 $ WAP013          : int  100 100 100 100 100 100 100 100 100 100 ...
 $ WAP014          : int  100 100 100 100 100 100 100 100 100 100 ...
 $ WAP015          : int  100 100 100 100 100 100 100 100 100 100 ...
 $ WAP016          : int  100 100 100 100 100 100 100 100 100 100 ...
 $ WAP017          : int  100 100 100 100 100 100 100 100 100 100 ...
 $ WAP018          : int  100 100 100 100 100 100 100 100 100 100 ...
 $ WAP019          : int  100 100 100 100 100 100 100 100 100 100 ...
 $ WAP020          : int  100 100 100 100 100 100 100 100 100 100 ...
 $ WAP021          : int  100 100 100 100 100 100 100 100 100 100 ...
 $ WAP022          : int  100 100 100 100 100 100 100 100 100 100 ...
 $ WAP023          : int  100 100 100 100 100 100 100 100 100 100 ...
 $ WAP024          : int  100 100 100 100 100 100 100 100 100 100 ...
 $ WAP025          : int  100 100 100 100 100 100 100 100 100 100 ...
 $ WAP026          : int  100 100 100 100 100 100 100 100 100 100 ...
 $ WAP027          : int  100 100 100 100 100 100 100 100 100 100 ...
 $ WAP028          : int  100 100 100 100 100 100 100 100 100 100 ...
 $ WAP029          : int  100 100 100 100 100 100 100 100 100 100 ...
 $ WAP030          : int  100 100 100 100 100 100 100 100 100 100 ...
 $ WAP031          : int  100 100 100 100 100 100 100 100 100 100 ...
 $ WAP032          : int  100 100 100 100 100 100 100 100 100 100 ...
...
 $ WAP520          : int  100 100 100 100 100 100 100 100 100 100 ...

I want to eliminate all WAP variables with zero variance.

Answer 1

Try this:

iszv <- grepl("WAP", names(Dtrain)) & sapply(Dtrain, is.numeric)
iszv[iszv] <- sapply(Dtrain[iszv], var, na.rm = TRUE) < 1e-9
Dtrain[,!iszv]
#    LONGITUDE LATITUDE FLOOR BUILDINGID SPACEID RELATIVEPOSITION USERID PHONEID  TIMESTAMP IDidentifier WAP008
# 1      -7541  4864921     2          1   L0031              L02      2      23 1371713733        L0400    100
# 2      -7537  4864934     2          1   L0031              L02      2      23 1371713691        L0400    100
# 3      -7519  4864950     2          1   L0028              L02      2      23 1371714095        L0394    -97
# 4      -7525  4864934     2          1   L0027              L02      2      23 1371713807        L0392    100
# 5      -7632  4864982     0          0   L0047              L02     11      13 1369909710        L0016    100
# 6         NA       NA     2          1   L0030              L02      2      23         NA         <NA>    100
# 7         NA       NA     2          1   L0028              L02      2      23         NA         <NA>    100
# 8         NA       NA     2          1   L0026              L02      2      23         NA         <NA>    100
# 9         NA       NA     2          1    <NA>              L02      2      23         NA         <NA>    100
# 10        NA       NA     2          1    <NA>                1      2      23         NA         <NA>    100

Though admittedly this is effectively the same as

Dtrain[,-(caret::nearZeroVar(Dtrain))]

---

Sample data:

Dtrain <- structure(list(LONGITUDE = c(-7541, -7537, -7519, -7525, -7632, NA, NA, NA, NA, NA), LATITUDE = c(4864921, 4864934, 4864950, 4864934, 4864982, NA, NA, NA, NA, NA), FLOOR = structure(c(2L, 2L, 2L, 2L, 1L, 2L, 2L, 2L, 2L, 2L), .Label = c("0", "2"), class = "factor"), BUILDINGID = structure(c(2L, 2L, 2L, 2L, 1L, 2L, 2L, 2L, 2L, 2L), .Label = c("0", "1"), class = "factor"), SPACEID = structure(c(5L, 5L, 3L, 2L, 6L, 4L, 3L, 1L, NA, NA), .Label = c("L0026", "L0027", "L0028", "L0030", "L0031", "L0047"), class = "factor"),      RELATIVEPOSITION = structure(c(2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 1L), .Label = c("1", "L02"), class = "factor"), USERID = c(2L, 2L, 2L, 2L, 11L, 2L, 2L, 2L, 2L, 2L), PHONEID = c(23L, 23L, 23L, 23L, 13L, 23L, 23L, 23L, 23L, 23L), TIMESTAMP = c(1371713733L, 1371713691L, 1371714095L, 1371713807L, 1369909710L, NA, NA, NA, NA, NA), IDidentifier = structure(c(4L, 4L, 3L, 2L, 1L, NA, NA, NA, NA, NA), .Label = c("L0016", "L0392", "L0394", "L0400"), class = "factor"), WAP001 = c(100L, 100L, 100L, 100L,      100L, 100L, 100L, 100L, 100L, 100L), WAP002 = c(100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L), WAP003 = c(100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L), WAP004 = c(100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L), WAP005 = c(100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L), WAP006 = c(100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L), WAP007 = c(100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L), WAP008 = c(100L, 100L, -97L,      100L, 100L, 100L, 100L, 100L, 100L, 100L), WAP009 = c(100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L), WAP010 = c(100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L), WAP011 = c(100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L), WAP012 = c(100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L), WAP013 = c(100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L), WAP014 = c(100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L), WAP015 = c(100L, 100L,      100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L), WAP016 = c(100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L), WAP017 = c(100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L), WAP018 = c(100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L), WAP019 = c(100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L), WAP020 = c(100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L), WAP021 = c(100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L), WAP022 = c(100L,      100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L), WAP023 = c(100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L), WAP024 = c(100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L), WAP025 = c(100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L), WAP026 = c(100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L), WAP027 = c(100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L), WAP028 = c(100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L), WAP029 = c(100L,      100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L), WAP030 = c(100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L), WAP031 = c(100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L), WAP032 = c(100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L, 100L)), class = "data.frame", row.names = c(NA, -10L))