to_histogram 1.0.4 → 1.0.5

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: bc64fe5ed589d7281499abede70e628d1d185fff
-  data.tar.gz: d0ed5fa267e3328cfd72fba7625ad500cfebc521
+  metadata.gz: aa1cb2265ca6164d3b0f7dd6292c1d499bfe0488
+  data.tar.gz: 7631a5d5fa1138dc8c606ea20796bd6728d8834b
 SHA512:
-  metadata.gz: 8062a66b610ee50b52c3053c4fb514cdb8edb70226b51500df6918e72dca7ad7457ca37e2826c0c73a0d2bb299390b6ff7481883f77e9114fbb0bf70c9e25dea
-  data.tar.gz: b4ad12cd120b69b86ce20ce032c03bb3e1c0c8112ac10f1783519ffe42bf176d3a0969cb47a0eeea24a869fc5145d20fe783ae4a949eed3fbd1e77e36b791b3c
+  metadata.gz: e5c41cd8733f140379ce4eadd116bd217549faef6e89ab6934374bae82a708def16b361015678b9d8fb896c6d5564577dd9ab84e3618e98e02b7708e59416c19
+  data.tar.gz: 426f7996e9e59fcccff17e8fa7d479b550b2f4367cb9ef6cfef45081dd780f5580f48260c4bb55c39c140d2045106aa380993742682796792ffcae094ca77df4

data/lib/bucketizer.rb

@@ -2,25 +2,35 @@ module ToHistogram
 
   class Bucketizer
 
-    def initialize(array, num_buckets)
+    def initialize(array, num_buckets, percentile)
       @arr                = array.sort
       @num_buckets        = num_buckets
-      @bucket_increments  = get_bucket_increment
+      @bucket_increments  = get_bucket_increment(percentile)
     end
     attr_reader :bucket_increments
 
     def create_buckets
       l_index               = 0
-      next_bucket           = (@bucket_increments == 1 && @arr[0] == 0) ? 0 : @bucket_increments
+      next_bucket           = @bucket_increments
       buckets               = []
 
+      # Deal with the special case where we have elements that == 0 and an increment sizes of 1 (count 0 as a value and don't lump it in with 1)
+      if(@arr.count(0) > 0 && next_bucket == 1)
+        bucket_0 = []
+        @arr.count(0).times { bucket_0 << @arr.shift }
+        buckets << bucket_0
+
+        next_bucket += 1
+      end
+
+      # Iterate thorough the remainder of the list in the normal case
       @arr.each_with_index do |e, i|
         break if buckets.length == (@num_buckets - 1)
 
-        if !(e <= next_bucket)
+        if (e >= next_bucket)
           buckets << @arr[l_index..(i - 1)]
 
-          # Special case where all of the results fit into the first bucket
+          # Special case here also where all of the results fit into the first bucket
           if buckets[0].length == @arr.length
             l_index = (@arr.length)
             break
@@ -40,13 +50,15 @@ module ToHistogram
     end
 
     private
-    def get_bucket_increment
+    def get_bucket_increment(percentile)
+      nth_percentile = percentile / 100.0
+
       if(@arr.length == 0)
         return 0
       elsif(@arr.length <= @num_buckets)
         increment = ((@arr[-1] - @arr[0]) / @num_buckets)
       else
-        increment = ((@arr[(@arr.length * 0.9).to_i - 1] - @arr[(@arr.length * 0.1).to_i - 1]) / @num_buckets.to_f).ceil
+        increment = ((@arr[(@arr.length * nth_percentile).to_i - 1] - @arr[0]) / @num_buckets.to_f).ceil
       end
 
       return increment

data/lib/histogram.rb

@@ -5,12 +5,14 @@ module ToHistogram
   class Histogram
     include Enumerable
 
-    def initialize(array, num_buckets)
-      bucketizer  = Bucketizer.new(array, num_buckets)
-      @buckets    = bucketizer.create_buckets
-      @increments = bucketizer.bucket_increments
+    def initialize(array, num_buckets, percentile)
+      bucketizer    = Bucketizer.new(array, num_buckets, percentile)
+      @buckets      = bucketizer.create_buckets
+      @increments   = bucketizer.bucket_increments
+      @percentile   = percentile
+      @num_buckets  = num_buckets
     end
-    attr_reader :increments
+    attr_reader :increments, :percentile, :num_buckets
 
     def each(&block)
       @buckets.each do |b|

data/lib/stdout_print.rb

@@ -1,16 +1,35 @@
 module ToHistogram
 
   class StdoutPrint
-    def initialize(histogram)
-      @histogram = histogram
+    def initialize(histogram, original_array)
+      @histogram      = histogram
+      @original_array = original_array.sort
     end
 
     def invoke
+      print_header
+      
+      print_body
+    end
+
+    private
+    def print_header
+      puts "\n**************************************************************"
+      puts "Results for #to_histogram(num_buckets: #{@histogram.num_buckets}, percentile: #{@histogram.percentile})"
+      puts "\n"
+      puts "Min Value: #{@original_array[0]}, Max Value: #{@original_array[-1]}"
+      puts "Mean: #{mean}, Median: #{median}, Mode: #{mode}"
+      puts "Histogram bucket sizes: #{@histogram.increments}"
+      puts "**************************************************************\n\n"
+    end
+
+    def print_body
       total_data_value_length = (@histogram.map { |b| b.length }).reduce(:+)
-      printf("%-20s %-20s %-30s %-20s \n\n", "range", "frequency", "  percentage (out of #{total_data_value_length})", "histogram (percetage)")
+      printf("%-20s %-20s %-30s %-20s \n\n", "Range", "Frequency", "  Percentage (out of #{total_data_value_length})", "Histogram (each * =~ 1%)")
       
       @histogram.each_with_index do |b, i|
-        range       = "#{b[0]} to #{b[-1]}"
+        next_bucket = (@histogram[i + 1]) ? @histogram[i + 1][0] : b[-1]
+        range       = "#{b[0]} to #{next_bucket}"
         frequency   = b.length
         percentage  = ((frequency.to_f / total_data_value_length) * 100)
         stars       = ''
@@ -27,5 +46,19 @@ module ToHistogram
       end
     end
 
+    def mean
+      @original_array.reduce(:+) / @original_array.length
+    end
+
+    def median
+      @original_array[@original_array.length / 2]
+    end
+
+    def mode
+      frequency = @original_array.inject(Hash.new(0)) { |h,v| h[v] += 1; h }
+      
+      @original_array.max_by { |v| frequency[v] }
+    end
+
   end
 end

data/lib/to_histogram.rb

@@ -2,9 +2,9 @@ require_relative './histogram'
 require_relative './stdout_print'
 
 class Array
-  def to_histogram(num_buckets: 10)
-    histogram       = ToHistogram::Histogram.new(self, num_buckets)    
-    stdout_print    = ToHistogram::StdoutPrint.new(histogram)
+  def to_histogram(num_buckets: 10, percentile: 100)
+    histogram       = ToHistogram::Histogram.new(self, num_buckets, percentile)    
+    stdout_print    = ToHistogram::StdoutPrint.new(histogram, self)
 
     stdout_print.invoke
 

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: to_histogram
 version: !ruby/object:Gem::Version
-  version: 1.0.4
+  version: 1.0.5
 platform: ruby
 authors:
 - Brett Sykes