Sum an array of hashes in Ruby

Question

I have a ruby array of 3 hashes. Each peace has information about report_data (consumption of 2 types of energy) and monthes_data (same for each). Please see the code below.

arr = [{:report_data=>
  [{:type=>
      {"id"=>1, "name"=>"electricity"},
    :data=>[10, 20, 30, 40]},
    {:type=>
      {"id"=>2, "name"=>"water"},
    :data=>[20, 30, 40, 50]}],
  :monthes_data=>
    {:monthes=>
      ["jan", "feb"]},

 {:report_data=>
  [{:type=>
      {"id"=>1, "name"=>"electricity"},
    :data=>[15, 25, 35, 45]},
    {:type=>
      {"id"=>2, "name"=>"water"},
    :data=>[25, 35, 45, 55]}],
  :monthes_data=>
    {:monthes=>
      ["jan", "feb"]},

 {:report_data=>
  [{:type=>
      {"id"=>1, "name"=>"electricity"},
    :data=>[17, 27, 37, 47]},
    {:type=>
      {"id"=>2, "name"=>"water"},
    :data=>[27, 37, 47, 57]}],
  :monthes_data=>
    {:monthes=>
      ["jan", "feb"]}]

I'm new to Ruby. Please help me to sum all the data by energy types. In the end I want to have one hash with report_data and monthes_data. I need the result look like:

{:report_data=>
  [{:type=>
      {:"id"=>1, "name"=>"electricity"},
    :data=>[42, 72, 102, 132]},
    {:type=>
      {"id"=>2, "name"=>"water"}},
    :data=>[72, 102, 132, 162]}],
  :monthes_data=>
    {:monthes=>
      ["jan", "feb"]}}

Answer 1

Code

def convert(arr)
  { :months_data=>arr.first[:months_data],
    :report_data=>arr.map { |h| h[:report_data] }.
      transpose.
      map { |d| { :type=>d.first[:type] }.
        merge(:data=>d.map { |g| g[:data] }.transpose.map { |a| a.reduce(:+) }) }
  }
end

Example

Half the battle in problems such as this one is visualizing the data. It's much clearer, imo, when written like this:

arr = [
  {:report_data=>[
     {:type=>{"id"=>1, "name"=>"electricity"}, :data=>[10, 20, 30, 40]},
     {:type=>{"id"=>2, "name"=>"water"},       :data=>[20, 30, 40, 50]}
     ],
   :months_data=>{:months=>["jan", "feb"]}
  },    
  {:report_data=>[
     {:type=>{"id"=>1, "name"=>"electricity"}, :data=>[15, 25, 35, 45]},
     {:type=>{"id"=>2, "name"=>"water"},       :data=>[25, 35, 45, 55]}
     ],
   :months_data=>{:months=>["jan", "feb"]}
  },    
  {:report_data=>[
     {:type=>{"id"=>1, "name"=>"electricity"}, :data=>[17, 27, 37, 47]},
     {:type=>{"id"=>2, "name"=>"water"},       :data=>[27, 37, 47, 57]}],
   :months_data=>{:months=>["jan", "feb"]}
  }
]

Let's try it:

convert(arr)
  #=> {:months_data=>{:months=>["jan", "feb"]},
  #    :report_data=>[
  #      {:type=>{"id"=>1, "name"=>"electricity"}, :data=>[42, 72, 102, 132]},
  #      {:type=>{"id"=>2, "name"=>"water"},       :data=>[72, 102, 132, 162]}
  #    ]
  #   }

Explanation

The first thing I did is concentrate on computing the sums, so I converted this to the values of :report_data. That key, and the key-value pair of months' data, which is the same for all elements (hashes) of arr, can be added back in later.

b = arr.map { |h| h[:report_data] }
  #=> [
  #    [{:type=>{"id"=>1, "name"=>"electricity"}, :data=>[10, 20, 30, 40]},
  #     {:type=>{"id"=>2, "name"=>"water"},       :data=>[20, 30, 40, 50]}
  #    ],
  #    [{:type=>{"id"=>1, "name"=>"electricity"}, :data=>[15, 25, 35, 45]},
  #     {:type=>{"id"=>2, "name"=>"water"},       :data=>[25, 35, 45, 55]}
  #    ],
  #    [{:type=>{"id"=>1, "name"=>"electricity"}, :data=>[17, 27, 37, 47]},
  #     {:type=>{"id"=>2, "name"=>"water"},       :data=>[27, 37, 47, 57]}
  #    ]
  #   ] 

If you are not certain that the elements of each array will be sorted by "id", you could write:

b = arr.map { |h| h[:report_data].sort_by { |g| g[:type]["id"] } }

c = b.transpose
  #=> [
  #    [{:type=>{"id"=>1, "name"=>"electricity"}, :data=>[10, 20, 30, 40]},
  #     {:type=>{"id"=>1, "name"=>"electricity"}, :data=>[15, 25, 35, 45]},
  #     {:type=>{"id"=>1, "name"=>"electricity"}, :data=>[17, 27, 37, 47]}
  #    ],
  #    [{:type=>{"id"=>2, "name"=>"water"},       :data=>[20, 30, 40, 50]},
  #     {:type=>{"id"=>2, "name"=>"water"},       :data=>[25, 35, 45, 55]},
  #     {:type=>{"id"=>2, "name"=>"water"},       :data=>[27, 37, 47, 57]}
  #    ]
  #   ] 

e = c.map {|d| { :type=>d.first[:type] }.
      merge(:data=>d.map { |g| g[:data] }.transpose.map { |a| a.reduce(:+) }) }
  #=> [{:type=>{"id"=>1, "name"=>"electricity"}, :data=>[42, 72, 102, 132]},
  #    {:type=>{"id"=>2, "name"=>"water"}      , :data=>[72, 102, 132, 162]}] 

Lastly, we need to put the put the key :report_data back in and add the months' data:

{ :months_data=>arr.first[:months_data], :report_data=>e }
  #=> {:months_data=>{:months=>["jan", "feb"]},
  #    :report_data=>[
  #      {:type=>{"id"=>1, "name"=>"electricity"}, :data=>[42, 72, 102, 132]},
  #      {:type=>{"id"=>2, "name"=>"water"},       :data=>[72, 102, 132, 162]}
  #    ]
  #   }

Answer 2

For clarity I've reformatted the input array and removed the :monthes_data key, since that seems to be unrelated to your question. Here's our data:

TL;DR

def zip_sum(arr1, arr2)
  return arr2 if arr1.nil?
  arr1.zip(arr2).map {|a, b| a + b }
end

def sum_report_data(arr)
  arr.flat_map do |item|
    item[:report_data].map {|datum| datum.values_at(:type, :data) }
  end
  .reduce({}) do |sums, (type, data)|
    sums.merge(type => data) do |_, old_data, new_data|
      zip_sum(old_data, new_data)
    end
  end
  .map {|type, data| { type: type, data: data } }
end

p sum_report_data(arr)
# =>
[ { type: { "id" => 1, "name" => "electricity" }, data: [ 42, 72, 102, 132 ] },
  { type: { "id" => 2, "name" => "water" }, data: [ 72, 102, 132, 162 ] }
]

Explanation

arr = [
  { report_data: [
      { type: { "id" => 1, "name" => "electricity" },
        data: [ 10, 20, 30, 40 ]
      },
      { type: { "id" => 2, "name" => "water" },
        data: [ 20, 30, 40, 50 ]
      }
    ]
  },

  { report_data: [
      { type: { "id" => 1, "name" => "electricity" },
        data: [ 15, 25, 35, 45 ]
      },
      { type: { "id" => 2, "name" => "water" },
        data: [ 25, 35, 45, 55 ]
      }
    ]
  },

  { report_data: [
      { type: { "id" => 1, "name" => "electricity" },
        data: [ 17, 27, 37, 47 ]
      },
      { type: { "id" => 2, "name" => "water" },
        data: [ 27, 37, 47, 57 ]
      }
    ]
  }
]

Step 1

First, let's define a helper method to sum the values of two arrays:

def zip_sum(arr1, arr2)
  return arr2 if arr1.nil?
  arr1.zip(arr2).map {|a, b| a + b }
end

zip_sum([ 1, 2, 3 ], [ 10, 20, 30 ])
# => [ 11, 22, 33 ]

zip_sum(nil, [ 5, 6, 7 ])
# => [ 5, 6, 7 ]

The way zip_sum works is by "zipping" the two arrays together using Enumerable#zip (e.g. [1, 2].zip([10, 20]) returns [ [1, 10], [2, 20] ]), then adding each pair together.

Step 2

Next, let's use Enumerable#flat_map to get the parts of the data we care about:

result1 = arr.flat_map do |item|
  item[:report_data].map {|datum| datum.values_at(:type, :data) }
end

# result1 =>
[ [ { "id" => 1, "name" => "electricity" }, [ 10, 20, 30, 40 ] ],
  [ { "id" => 2, "name" => "water" },       [ 20, 30, 40, 50 ] ],
  [ { "id" => 1, "name" => "electricity" }, [ 15, 25, 35, 45 ] ],
  [ { "id" => 2, "name" => "water" },       [ 25, 35, 45, 55 ] ],
  [ { "id" => 1, "name" => "electricity" }, [ 17, 27, 37, 47 ] ],
  [ { "id" => 2, "name" => "water" },       [ 27, 37, 47, 57 ] ]
]

Above we've just grabbed the :type and :data values out of each hash the :report_data arrays.

Step 3

Next let's use Enumerable#reduce to iterate over the array of arrays and calculate a running sum of the :data values using the zip_sum method we defined earlier:

result2 = result1.reduce({}) do |sums, (type, data)|
  sums.merge(type => data) do |_, old_data, new_data|
    zip_sum(old_data, new_data)
  end
end

# result2 =>
{ { "id" => 1, "name" => "electricity" } => [ 42,  72, 102, 132 ],
  { "id" => 2, "name" => "water" }       => [ 72, 102, 132, 162 ]
}

The result might look a little odd to you because we usually use strings or symbols as hash keys, but in this hash we're using other hashes (the :type values from above) as keys. That's one nice thing about Ruby: You can use any object as a key in a hash.

Inside the reduce block, sums is the hash that's ultimately returned. It starts out as an empty hash ({}, the value we passed to reduce as an argument). type is the hash we're using as a key and data is the array of integers. In each iteration the next values from the result2 array are assigned to type, but sums is updated with whatever value was returned at the end of the block in the previous iteration.

We're using Hash#merge in kind of a tricky way:

sums.merge(type => data) do |_, old_data, new_data|
  zip_sum(old_data, new_data)
end

This merges the hash { type => data } (remember that type is the :type hash and data is the array of integers) into the hash sums. If there are any key collisions, the block will be invoked. Since we only have one key, type, then the block will be invoked if sums[type] already exists. If it does, we call zip_sum with the previous value of sums[type] and data, effectively keeping a running sum of data.

In effect, it's basically doing this:

sums = {}

type, data = result2[0]
sums[type] = zip_sum(sums[type], data)

type, data = result2[1]
sums[type] = zip_sum(sums[type], data)

type, data = result2[3]
# ...and so on.

Step 4

We now have this hash in result3:

{ { "id" => 1, "name" => "electricity" } => [ 42,  72, 102, 132 ],
  { "id" => 2, "name" => "water" }       => [ 72, 102, 132, 162 ]
}

That's the data we want, so now we just have to take it out of this weird format and put it into a regular hash with the keys :type and :data:

result3 = result2.map {|type, data| { type: type, data: data } }

# result3 =>
[ { type: { "id" => 1, "name" => "electricity" },
    data: [ 42, 72, 102, 132 ]
  },
  { type: { "id" => 2, "name" => "water" },
    data: [ 72, 102, 132, 162 ]
  }
]

Answer 3

arr = [{:report_data=>
  [{:type=>
      {"id"=>1, "name"=>"electricity"},
    :data=>[10, 20, 30, 40]},
    {:type=>
      {"id"=>2, "name"=>"water"},
    :data=>[20, 30, 40, 50]}],
  :monthes_data=>
    {:monthes=>
      ["jan", "feb"]}},

 {:report_data=>
  [{:type=>
      {"id"=>1, "name"=>"electricity"},
    :data=>[15, 25, 35, 45]},
    {:type=>
      {"id"=>2, "name"=>"water"},
    :data=>[25, 35, 45, 55]}],
  :monthes_data=>
    {:monthes=>
      ["jan", "feb"]}},

 {:report_data=>
  [{:type=>
      {"id"=>1, "name"=>"electricity"},
    :data=>[17, 27, 37, 47]},
    {:type=>
      {"id"=>2, "name"=>"water"},
    :data=>[27, 37, 47, 57]}],
  :monthes_data=>
    {:monthes=>
      ["jan", "feb"]}}]


acc = {}
arr.each do
    |e| e[:report_data].each_with_index do
        |e, idx| 
        type = e[:type]['id']
        e[:data].each_with_index do 
            |e, idx|
            acc[type] = [] if not acc[type]
            acc[type][idx] = (acc[type][idx] or 0) + e  
        end
    end
end
p acc

outputs

{1=>[42, 72, 102, 132], 2=>[72, 102, 132, 162]}

You should be able to reformat this into your record