Only push specific substring to array in Ruby

first_array.map{|s| s[/\d+/].to_i}
# => [8, 7] 

And I have to add this super short but complicated one-liner solution:

a = ["Promoter: 8", "Passive: 7"]
p a.grep(/(\d+)/){$&.to_i} #=> [8,7]

Your question, as formulated, has an easy practical answer, already provided by others. But it seems to me, that your array of strings

a = ["Promoter: 8", "Passive: 7"]

envies being a Hash. So, from broader perspective, I would take freedom of converting it to a Hash first:

require 'pyper' # (type "gem install pyper" in your command line to install it)
hsh = Hash[ a.τBmm2dτ &/(\w+): *(\d+)/.method( :match ) ]
#=> {"Promoter"=>"8", "Passive"=>"7"}
# (The construction of #τBmm2dτ Pyper method will be explained in the appendix.)

Now, having your input data in a hash, you can do things with them more easily, eg.

hsh.τmbtiτ
#=> [8, 7]

APPENDIX: Explanation of the Pyper methods.

Pyper methods are similar to Lisp #car/#cdr methods in that, that a combination of letters controls the method behavior. In the first method, #τBmm2dτ:

• τ - opening and ending character
• m - means #map
• B - means take a block
• 2 - means first 3 elements of an array
• d - means all elements except the first one (same meaning as in #cdr, btw.)

So, in #τBmm2dτ, Bm applies the block as follows:

x = ["Promoter: 8", "Passive: 7"].map  &/(\w+): *(\d+)/.method( :match )
#=> [#<MatchData "Promoter: 8" 1:"Promoter" 2:"8">, #<MatchData "Passive: 7" 1:"Passive" 2:"7">]
# Which results in an array of 2 MatchData objects.

Then, m2d chars map (m) the MatchData objects using 2 and d chars. Character 2 gives

x = x.map { |e| e.to_a.take 3 }
#=> [["Promoter: 8", "Promoter", "8"], ["Passive: 7", "Passive", "7"]]

and d removes the first element from each:

x = x.map { |e| e.drop 1 }
#=> [["Promoter", "8"], ["Passive", "7"]]

In the secon method, #τmbtiτ, m means again #map, b means take the second element, and ti means convert it to Integer:

{"Promoter"=>"8", "Passive"=>"7"}.to_a.map { |e| Integer e[1] }
#=> [8, 7]

If the integer part of each string in (which look like members of a hash) is always preceded by at least one space, and there is no other whitespace (other than possibly at the beginning of the string), you could do this:

first_array = ["Promoter: 8", "Passive: 7"]
Hash[*first_array.map(&:split).flatten].values.map(&:to_i) # => [8,7]

• map first_array => [["Promoter:", "8"], ["Passive:", "7"]]
• flatten => ["Promoter:", "8", "Passive:", "7"]
• convert to hash => {"Promoter:" => "8", "Passive:" => "7"}
• get hash values => ["8", "7"]
• convert to ints => [8, 7]

Note the need for the splat:

 Hash[*["Promoter:", "8", "Passive:", "7"]]
   => Hash["Promoter:", "8", "Passive:", "7"]                                               
   => {"Promoter:" => "8", "Passive:" => "7"}