R – Sort rows in data.table in decreasing order on string key `order(-x,v)` gives error on data.table 1.9.4 or earlier

data.tablekeyrsortingstring

Let's say I have the following data.table in R:

  library(data.table)
  DT = data.table(x=rep(c("b","a","c"),each=3), y=c(1,3,6), v=1:9)

I want to order it by two columns (say columns x and v). I used this:

 DT[order(x,v)] # sorts first by x then by v (both in ascending order)

But now, I want to sort it by x (in decreasing order) and have the following code:

  DT[order(-x)] #Error in -x : invalid argument to unary operator

Therefore, I think this error is due to the fact that class(DT$x)=character. Could you give me any suggestion in order to solve this issue?

I know I can use DT[order(x,decreasing=TRUE)], but I want to know the syntax to sort by several columns using both ways (some decreasing, some increasing) at the same time.

Note that if you use DT[order(-y,v)] the result is ok, but if you use DT[order(-x,v)] there is an error. So, my question is: how to solve this error?

Best Answer

Update

data.table v1.9.6+ now supports OP's original attempt and the following answer is no longer necessary.

You can use DT[order(-rank(x), y)].

   x y v
1: c 1 7
2: c 3 8
3: c 6 9
4: b 1 1
5: b 3 2
6: b 6 3
7: a 1 4
8: a 3 5
9: a 6 6

Related Solutions

C++ – Static constant string (class member)

You have to define your static member outside the class definition and provide the initializer there.

First

// In a header file (if it is in a header file in your case)
class A {   
private:      
  static const string RECTANGLE;
};

and then

// In one of the implementation files
const string A::RECTANGLE = "rectangle";

The syntax you were originally trying to use (initializer inside class definition) is only allowed with integral and enum types.

Starting from C++17 you have another option, which is quite similar to your original declaration: inline variables

// In a header file (if it is in a header file in your case)
class A {   
private:      
  inline static const string RECTANGLE = "rectangle";
};

No additional definition is needed.

Starting from C++20 instead of const you can declare it constexpr in this variant. Explicit inline would no longer be necessary, since constexpr implies inline.

Ruby – How to sort an array in descending order in Ruby

It's always enlightening to do a benchmark on the various suggested answers. Here's what I found out:

#!/usr/bin/ruby

require 'benchmark'

ary = []
1000.times { 
  ary << {:bar => rand(1000)} 
}

n = 500
Benchmark.bm(20) do |x|
  x.report("sort")               { n.times { ary.sort{ |a,b| b[:bar] <=> a[:bar] } } }
  x.report("sort reverse")       { n.times { ary.sort{ |a,b| a[:bar] <=> b[:bar] }.reverse } }
  x.report("sort_by -a[:bar]")   { n.times { ary.sort_by{ |a| -a[:bar] } } }
  x.report("sort_by a[:bar]*-1") { n.times { ary.sort_by{ |a| a[:bar]*-1 } } }
  x.report("sort_by.reverse!")   { n.times { ary.sort_by{ |a| a[:bar] }.reverse } }
end

                          user     system      total        real
sort                  3.960000   0.010000   3.970000 (  3.990886)
sort reverse          4.040000   0.000000   4.040000 (  4.038849)
sort_by -a[:bar]      0.690000   0.000000   0.690000 (  0.692080)
sort_by a[:bar]*-1    0.700000   0.000000   0.700000 (  0.699735)
sort_by.reverse!      0.650000   0.000000   0.650000 (  0.654447)

I think it's interesting that @Pablo's sort_by{...}.reverse! is fastest. Before running the test I thought it would be slower than "-a[:bar]" but negating the value turns out to take longer than it does to reverse the entire array in one pass. It's not much of a difference, but every little speed-up helps.

Please note that these results are different in Ruby 1.9

Here are results for Ruby 1.9.3p194 (2012-04-20 revision 35410) [x86_64-darwin10.8.0]:

                           user     system      total        real
sort                   1.340000   0.010000   1.350000 (  1.346331)
sort reverse           1.300000   0.000000   1.300000 (  1.310446)
sort_by -a[:bar]       0.430000   0.000000   0.430000 (  0.429606)
sort_by a[:bar]*-1     0.420000   0.000000   0.420000 (  0.414383)
sort_by.reverse!       0.400000   0.000000   0.400000 (  0.401275)

These are on an old MacBook Pro. Newer, or faster machines, will have lower values, but the relative differences will remain.

Here's a bit updated version on newer hardware and the 2.1.1 version of Ruby:

#!/usr/bin/ruby

require 'benchmark'

puts "Running Ruby #{RUBY_VERSION}"

ary = []
1000.times {
  ary << {:bar => rand(1000)}
}

n = 500

puts "n=#{n}"
Benchmark.bm(20) do |x|
  x.report("sort")               { n.times { ary.dup.sort{ |a,b| b[:bar] <=> a[:bar] } } }
  x.report("sort reverse")       { n.times { ary.dup.sort{ |a,b| a[:bar] <=> b[:bar] }.reverse } }
  x.report("sort_by -a[:bar]")   { n.times { ary.dup.sort_by{ |a| -a[:bar] } } }
  x.report("sort_by a[:bar]*-1") { n.times { ary.dup.sort_by{ |a| a[:bar]*-1 } } }
  x.report("sort_by.reverse")    { n.times { ary.dup.sort_by{ |a| a[:bar] }.reverse } }
  x.report("sort_by.reverse!")   { n.times { ary.dup.sort_by{ |a| a[:bar] }.reverse! } }
end

# >> Running Ruby 2.1.1
# >> n=500
# >>                            user     system      total        real
# >> sort                   0.670000   0.000000   0.670000 (  0.667754)
# >> sort reverse           0.650000   0.000000   0.650000 (  0.655582)
# >> sort_by -a[:bar]       0.260000   0.010000   0.270000 (  0.255919)
# >> sort_by a[:bar]*-1     0.250000   0.000000   0.250000 (  0.258924)
# >> sort_by.reverse        0.250000   0.000000   0.250000 (  0.245179)
# >> sort_by.reverse!       0.240000   0.000000   0.240000 (  0.242340)

New results running the above code using Ruby 2.2.1 on a more recent Macbook Pro. Again, the exact numbers aren't important, it's their relationships:

Running Ruby 2.2.1
n=500
                           user     system      total        real
sort                   0.650000   0.000000   0.650000 (  0.653191)
sort reverse           0.650000   0.000000   0.650000 (  0.648761)
sort_by -a[:bar]       0.240000   0.010000   0.250000 (  0.245193)
sort_by a[:bar]*-1     0.240000   0.000000   0.240000 (  0.240541)
sort_by.reverse        0.230000   0.000000   0.230000 (  0.228571)
sort_by.reverse!       0.230000   0.000000   0.230000 (  0.230040)

Updated for Ruby 2.7.1 on a Mid-2015 MacBook Pro:

Running Ruby 2.7.1
n=500     
                           user     system      total        real
sort                   0.494707   0.003662   0.498369 (  0.501064)
sort reverse           0.480181   0.005186   0.485367 (  0.487972)
sort_by -a[:bar]       0.121521   0.003781   0.125302 (  0.126557)
sort_by a[:bar]*-1     0.115097   0.003931   0.119028 (  0.122991)
sort_by.reverse        0.110459   0.003414   0.113873 (  0.114443)
sort_by.reverse!       0.108997   0.001631   0.110628 (  0.111532)

...the reverse method doesn't actually return a reversed array - it returns an enumerator that just starts at the end and works backwards.

The source for Array#reverse is:

               static VALUE
rb_ary_reverse_m(VALUE ary)
{
    long len = RARRAY_LEN(ary);
    VALUE dup = rb_ary_new2(len);

    if (len > 0) {
        const VALUE *p1 = RARRAY_CONST_PTR_TRANSIENT(ary);
        VALUE *p2 = (VALUE *)RARRAY_CONST_PTR_TRANSIENT(dup) + len - 1;
        do *p2-- = *p1++; while (--len > 0);
    }
    ARY_SET_LEN(dup, RARRAY_LEN(ary));
    return dup;
}

do *p2-- = *p1++; while (--len > 0); is copying the pointers to the elements in reverse order if I remember my C correctly, so the array is reversed.

Best Answer

Update

Related Solutions

C++ – Static constant string (class member)

Ruby – How to sort an array in descending order in Ruby

Related Topic