Weird behaviour by ordering a data frame

Question

I have the following data frame that I want to order by the fifth column ("Distance"). When I try `

df.order <- df[order(df[, 5]), ]

I always get the following error message.

Error in order(df[, 5]) : unimplemented type 'list' in 'orderVector1'`

I don't know why R consider my data frame as a list. Running is.data.frame(df) returns TRUE. I have to admit that is.list(df) also returns TRUE. Is is possible to force my data frame to be only a data frame and not a list? Thanks for your help.

structure(list(ID = list(1, 2, 3, 4, 5, 6, 7, 8, 9, 10), 
               Latitude = list(50.7368, 50.7368, 50.7368, 50.7369, 50.7369, 50.737, 50.737, 50.7371, 50.7371, 50.7371), 
               Longitude = list(6.0873, 6.0873, 6.0873, 6.0872, 6.0872, 6.0872, 6.0872, 6.0872, 6.0872, 6.0872), 
               Elevation = list(269.26, 268.99, 268.73, 268.69, 268.14, 267.87, 267.61, 267.31, 267.21, 267.02), 
               Distance = list(119.4396, 119.4396, 119.4396, 121.199, 121.199, 117.5658, 117.5658, 114.9003, 114.9003, 114.9003), 
               RxPower = list(-52.6695443922406, -52.269130891243, -52.9735258244422, -52.2116571930007, -51.7784534281727, -52.7703448813654, -51.6558862949081, -52.2892907635308, -51.8322993596551, -52.4971436682333)), 
          .Names = c("ID", "Latitude", "Longitude", "Elevation", "Distance", "RxPower"),
          row.names = c(NA, 10L), class = "data.frame")

Answer 1

Your data frame contains lists, not vectors. You can convert this data frame to the "classical" format using as.data.frame and unlist:

df2 <- as.data.frame(lapply(df, unlist))

Now, the new data frame could be sorted in the intended way:

df2[order(df2[, 5]), ]

Answer 2

I've illustrated with a small example what's the problem:

df <- structure(list(ID = c(1, 2, 3, 4), 
          Latitude = c(50.7368, 50.7368, 50.7368, 50.7369), 
          Longitude = c(6.0873, 6.0873, 6.0873, 6.0872), 
          Elevation = c(269.26, 268.99, 268.73, 268.69), 
          Distance = c(119.4396, 119.4396, 119.4396, 121.199), 
          RxPower = c(-52.6695443922406, -52.269130891243, -52.9735258244422, 
                         -52.2116571930007)), 
          .Names = c("ID", "Latitude", "Longitude", "Elevation", "Distance", "RxPower"), 
          row.names = c(NA, 4L), class = "data.frame")

Notice that list only occurs once. And all the values are wrapped by c(.) and not list(.). This is why doing sapply(df, class) on your data resulted in all columns having class list.

Now,

> sapply(df, classs)
#       ID  Latitude Longitude Elevation  Distance   RxPower 
# "numeric" "numeric" "numeric" "numeric" "numeric" "numeric" 

Now order works:

> df[order(df[,4]), ]  
#   ID Latitude Longitude Elevation Distance   RxPower
# 4  4  50.7369    6.0872    268.69 121.1990 -52.21166
# 3  3  50.7368    6.0873    268.73 119.4396 -52.97353
# 2  2  50.7368    6.0873    268.99 119.4396 -52.26913
# 1  1  50.7368    6.0873    269.26 119.4396 -52.66954

Answer 3

This turns you data.frame of lists into a matrix:

mat <- sapply(df,unlist)

Now you can order it.

mat[order(mat[,5]),]

If all columns are of one type, e.g., numeric, a matrix often is preferable, because operations on matrices are faster than on data.frames. However, you can transform to a data.frame using as.data.frame(mat).

Btw, a data.frame is a special kind of list and thus is.list returns TRUE for every data.frame.