# Counting Frequency of Values in an Array (And Sorting by Frequency?)

Discussion in 'Ruby' started by x1, Oct 12, 2006.

1. ### x1Guest

Is there no method for an array that will tell me the # of occurrences
for an item?

IE: ["a", "a", "a", "b", "c", "c"].count("a") #producing 3 ?

I almost thought that rindex would do the trick when looking at the
class docs but.. the example was just engineered to trick me :-(

I realize I could pass these to a block and count but.. wanted to make
sure it didn't exist. If not, why? Thank you.. ( I did search btw.. no
avail )

Also, what's the best way of printing out each unique item and the
number of times it occurs, sorted by numerically by the number of
times it occurs?

IE: in my example above, i'd like to see (sorted by occurrence
greatest to least)
#desired output:
a: 3
c: 2
b: 1

Or sorted from least to greatest:
#desired output:
b: 1
c: 2
a: 3

I was able to hack it by using a hash doing various things to it.. but
it didn't seem "rubyish".

Thank you for any input.

x1, Oct 12, 2006

2. ### PhrogzGuest

x1 wrote:
> I was able to hack it by using a hash doing various things to it.. but
> it didn't seem "rubyish".

What could be more ruby-ish than monkeypatching a built-in class and
using inject in the process?

class Array
def counts
inject( Hash.new(0) ){ |hash,element|
hash[ element ] +=1
hash
}
end
def counts_up
counts.sort_by{ |k,v| v }
end
def counts_down
counts.sort_by{ |k,v| -v }
end
end
a = ["a", "a", "a", "b", "c", "c"]
p a.counts, a.counts_up, a.counts_down
#=> {"a"=>3, "b"=>1, "c"=>2}
#=> [["b", 1], ["c", 2], ["a", 3]]
#=> [["a", 3], ["c", 2], ["b", 1]]

Phrogz, Oct 12, 2006

3. ### Logan CapaldoGuest

Re: Counting Frequency of Values in an Array (And Sorting byFrequency?)

On Thu, Oct 12, 2006 at 11:52:19AM +0900, x1 wrote:
> Is there no method for an array that will tell me the # of occurrences
> for an item?
>
> IE: ["a", "a", "a", "b", "c", "c"].count("a") #producing 3 ?
>

array.select { |i| i == 'a' }.length
or array.inject(0) { |count, item| count += 1 if item == 'a'; count }

> I almost thought that rindex would do the trick when looking at the
> class docs but.. the example was just engineered to trick me :-(
>
> I realize I could pass these to a block and count but.. wanted to make
> sure it didn't exist. If not, why? Thank you.. ( I did search btw.. no
> avail )
>
> Also, what's the best way of printing out each unique item and the
> number of times it occurs, sorted by numerically by the number of
> times it occurs?
>
> IE: in my example above, i'd like to see (sorted by occurrence
> greatest to least)
> #desired output:
> a: 3
> c: 2
> b: 1
>
>
> Or sorted from least to greatest:
> #desired output:
> b: 1
> c: 2
> a: 3
>
> I was able to hack it by using a hash doing various things to it.. but
> it didn't seem "rubyish".
>

As you say:
puts array.inject(Hash.new(0)) { |hash, item| hash[item] += 1
hash }.sort_by { |k, v| v }.map { |k, v| "#{k}:#{v}" }
> Thank you for any input.

Logan Capaldo, Oct 12, 2006

Hi,

At Thu, 12 Oct 2006 11:52:19 +0900,
x1 wrote in [ruby-talk:219218]:
> Is there no method for an array that will tell me the # of occurrences
> for an item?
>
> IE: ["a", "a", "a", "b", "c", "c"].count("a") #producing 3 ?

FYI, Enumerable in 1.9 has that method.

--

5. ### x1Guest

Wow Capaldo, that worked! I fear for the next java programmer who has
to make sense of my code when I leave ;-)

To reverse sort, I added .reverse.. Here's the final product:
puts ["a", "a", "a", "b", "c", "c"].inject(Hash.new(0)) { |hash, item|
hash[item] += 1
hash }.sort_by { |k, v| v }.reverse.map { |k, v| "#{k}:#{v}" }

Thanks so much.

Nakada, I look forward to using it in 1.9

> Hi,
>
> At Thu, 12 Oct 2006 11:52:19 +0900,
> x1 wrote in [ruby-talk:219218]:
> > Is there no method for an array that will tell me the # of occurrences
> > for an item?
> >
> > IE: ["a", "a", "a", "b", "c", "c"].count("a") #producing 3 ?

>
> FYI, Enumerable in 1.9 has that method.
>
> --
>
>

x1, Oct 12, 2006
6. ### x1Guest

and.. so you're aware.. my hacky code was something like this:

items = {}
["a", "a", "a", "b", "c", "c"].each do |i|
if items.include? i
items += 1
else
items = 1
end
end

items.sort {|a,b| a[1]<=>b[1]}.reverse.each do |a, b|
puts a + ":" + b.to_s
end

horrific eh?

On 10/11/06, x1 <> wrote:
> Wow Capaldo, that worked! I fear for the next java programmer who has
> to make sense of my code when I leave ;-)
>
> To reverse sort, I added .reverse.. Here's the final product:
> puts ["a", "a", "a", "b", "c", "c"].inject(Hash.new(0)) { |hash, item|
> hash[item] += 1
> hash }.sort_by { |k, v| v }.reverse.map { |k, v| "#{k}:#{v}" }
>
> Thanks so much.
>
> Nakada, I look forward to using it in 1.9
>
> > Hi,
> >
> > At Thu, 12 Oct 2006 11:52:19 +0900,
> > x1 wrote in [ruby-talk:219218]:
> > > Is there no method for an array that will tell me the # of occurrences
> > > for an item?
> > >
> > > IE: ["a", "a", "a", "b", "c", "c"].count("a") #producing 3 ?

> >
> > FYI, Enumerable in 1.9 has that method.
> >
> > --
> >
> >

>

x1, Oct 12, 2006
7. ### Daniel SheppardGuest

items =3D Hash.new(0)
["a","a","a","b","c","c"].each do |i|
items +=3D 1
end
items.sort_by {|key,value| -value}.each do |key, value|
puts "#{a}:#{b}"
end

Slightly less horrific?

> and.. so you're aware.. my hacky code was something like this:
>=20
> items =3D {}
> ["a", "a", "a", "b", "c", "c"].each do |i|
> if items.include? i
> items +=3D 1
> else
> items =3D 1
> end
> end
>=20
> items.sort {|a,b| a[1]<=3D>b[1]}.reverse.each do |a, b|
> puts a + ":" + b.to_s
> end
>=20
>=20
> horrific eh?

Daniel Sheppard, Oct 12, 2006
8. ### Park HeesobGuest

Hi,
>From: x1 <>
>To: (ruby-talk ML)
>Subject: Counting Frequency of Values in an Array (And Sorting by
>Frequency?)
>Date: Thu, 12 Oct 2006 11:52:19 +0900
>
>Is there no method for an array that will tell me the # of occurrences
>for an item?
>
>IE: ["a", "a", "a", "b", "c", "c"].count("a") #producing 3 ?
>

You can use ["a", "a", "a", "b", "c", "c"].grep("a").size

>I almost thought that rindex would do the trick when looking at the
>class docs but.. the example was just engineered to trick me :-(
>
>I realize I could pass these to a block and count but.. wanted to make
>sure it didn't exist. If not, why? Thank you.. ( I did search btw.. no
>avail )
>
>Also, what's the best way of printing out each unique item and the
>number of times it occurs, sorted by numerically by the number of
>times it occurs?
>
>IE: in my example above, i'd like to see (sorted by occurrence
>greatest to least)
>#desired output:
>a: 3
>c: 2
>b: 1
>

array.uniq.sort_by{|x|array.grep(x).size}.reverse.each{|x|puts "#{x}:
#{array.grep(x).size}"}

>
>Or sorted from least to greatest:
>#desired output:
>b: 1
>c: 2
>a: 3
>

array.uniq.sort_by{|x|array.grep(x).size}.each{|x|puts "#{x}:
#{array.grep(x).size}"}

Regards,
Park Heesob

Park Heesob, Oct 12, 2006
9. ### x1Guest

Ah ok.. I'm with ya. Meet "sort_by", my new friend. Thanks again :-D

On 10/12/06, Daniel Sheppard <> wrote:
> items = Hash.new(0)
> ["a","a","a","b","c","c"].each do |i|
> items += 1
> end
> items.sort_by {|key,value| -value}.each do |key, value|
> puts "#{a}:#{b}"
> end
>
> Slightly less horrific?
>
> > and.. so you're aware.. my hacky code was something like this:
> >
> > items = {}
> > ["a", "a", "a", "b", "c", "c"].each do |i|
> > if items.include? i
> > items += 1
> > else
> > items = 1
> > end
> > end
> >
> > items.sort {|a,b| a[1]<=>b[1]}.reverse.each do |a, b|
> > puts a + ":" + b.to_s
> > end
> >
> >
> > horrific eh?

>
>
>

x1, Oct 12, 2006
10. ### Rick DeNataleGuest

On 10/11/06, x1 <> wrote:
> Is there no method for an array that will tell me the # of occurrences
> for an item?
>
> IE: ["a", "a", "a", "b", "c", "c"].count("a") #producing 3 ?
>

Smalltalk has a collection class called Bag, which is an unordered
collection of objects which keeps track of the number of occurences of
each equal element in the collection, so you can do something like:

bag <- Bag.new
bag.occurencesOf: 1 => 2
bag.occurencesOf: 1 => 1
bag.remove: 1
bag.occurrencesOf: 1 => 1

Now Bag is kind of the black sheep of the Smalltalk collection
classes. Most Smalltalkers would either never use it or overuse it.
The only use I could think of was to implement a histogram.

--
Rick DeNatale

My blog on Ruby