statsample 0.17.0 → 0.18.0

data/lib/statsample/graph.rb

@@ -2,6 +2,10 @@ require 'statsample/graph/scatterplot'
 require 'statsample/graph/boxplot'
 require 'statsample/graph/histogram'
 module Statsample
+  # Several Graph, based on Rubyvis[http://rubyvis.rubyforge.org/]
+  # * Statsample::Graph::Boxplot
+  # * Statsample::Graph::Histogram  
+  # * Statsample::Graph::Scatterplot
   module Graph
   end
 end

data/lib/statsample/graph/boxplot.rb

@@ -45,6 +45,10 @@ module Statsample
       attr_accessor :maximum
       # Vectors to box-ploting
       attr_accessor :vectors
+      # The rotation angle, in radians. Text is rotated clockwise relative 
+      # to the anchor location. For example, with the default left alignment, 
+      # an angle of Math.PI / 2 causes text to proceed downwards. The default angle is zero.      
+      attr_accessor :label_angle
       
       attr_reader :x_scale, :y_scale
       # Create a new Boxplot.
@@ -66,7 +70,8 @@ module Statsample
           :margin_left=>20,
           :margin_right=>20,
           :minimum=>nil,
-          :maximum=>nil
+          :maximum=>nil,
+          :label_angle=>0
         }
         @opts=opts_default.merge(opts)
         opts_default.keys.each {|k| send("#{k}=", @opts[k]) }
@@ -94,26 +99,24 @@ module Statsample
         
         data=@vectors.map {|v|
           out={:percentil_25=>v.percentil(25), :median=>v.median, :percentil_75=>v.percentil(75), :name=>v.name}
-          out[:iqr]=out[:percentil_75]-out[:percentil_25]
+          out[:iqr]=out[:percentil_75] - out[:percentil_25]
           
-          irq_max=out[:percentil_75]+out[:iqr]
-          irq_min=out[:percentil_25]-out[:iqr]
+          irq_max=out[:percentil_75] + out[:iqr]
+          irq_min=out[:percentil_25] - out[:iqr]
           
           # Find the last data inside the margin
-          min=out[:percentil_25]
-          max=out[:percentil_75]
+          min = out[:percentil_25]
+          max = out[:percentil_75]
           
           v.each {|d|
-            min=d if d<min and d>irq_min
-            max=d if d>max and d<irq_max
+            min=d if d < min and d > irq_min
+            max=d if d > max and d < irq_max
           }
           # Whiskers!
           out[:low_whisker]=min
           out[:high_whisker]=max
           # And now, data outside whiskers
-          out[:outliers]=v.data_with_nils.find_all {|d| 
-            d<min or d>max
-          }
+          out[:outliers]=v.data_with_nils.find_all {|d| d < min or d > max }
           out
         }
         
@@ -132,7 +135,6 @@ module Statsample
             bottom y_scale
             stroke_style {|d| d!=0 ? "#eee" : "#000"}
             label(:anchor=>'left') do
-              visible {|d| true}
               text y_scale.tick_format
             end
           end
@@ -140,23 +142,27 @@ module Statsample
             bottom 0
             stroke_style 'black'
           end
+          
+          # Labels
+          
           pan.label  do |l|
             l.data data
-            l.left  {|v|  x_scale.scale(index)}
-            l.bottom -15
+            l.text_angle that.label_angle
+            l.left  {|v| x_scale[index] }
+            l.bottom(-15)
             l.text {|v,x| v[:name]}
           end
           
           pan.panel do |bp|
             bp.data data
-            bp.left {|v|  x_scale.scale(index)}
+            bp.left {|v|  x_scale[index]}
             bp.width x_scale.range_band
             
             
             # Bar
             bp.bar do |b|
-              b.bottom {|v| y_scale.scale(v[:percentil_25])}
-              b.height {|v| y_scale.scale(v[:percentil_75]) - y_scale.scale(v[:percentil_25]) }
+              b.bottom {|v| y_scale[v[:percentil_25]]}
+              b.height {|v| y_scale[v[:percentil_75]] - y_scale[v[:percentil_25]] }
               b.line_width 1
               b.stroke_style  {|v| 
                 if that.groups
@@ -177,34 +183,39 @@ module Statsample
             end
             # Median
             bp.rule do |r|
-              r.bottom {|v| y_scale.scale(v[:median])}
+              r.bottom {|v| y_scale[v[:median]]}
               r.width x_scale.range_band
               r.line_width 2
             end
-            
+            ##
             # Whiskeys
+            ##
+            # Low whiskey
             bp.rule do |r|
-              r.visible {|v| v[:percentil_25]>v[:low_whisker]}
-              r.bottom {|v| y_scale.scale(v[:low_whisker])}              
+              r.visible {|v| v[:percentil_25] > v[:low_whisker]}
+              r.bottom {|v| y_scale[v[:low_whisker]]}              
             end
+            
             bp.rule do |r|
-              r.visible {|v| v[:percentil_25]>v[:low_whisker]}
-              r.bottom {|v| y_scale.scale(v[:low_whisker])}              
+              r.visible {|v| v[:percentil_25] > v[:low_whisker]}
+              r.bottom {|v| y_scale[v[:low_whisker]]}              
               r.left {|v| x_scale.range_band / 2.0}
-              r.height {|v| y_scale.scale(v[:percentil_25])-y_scale.scale(v[:low_whisker])}
+              r.height {|v| y_scale.scale(v[:percentil_25]) - y_scale.scale(v[:low_whisker])}
             end
+            # High whiskey
+
             bp.rule do |r|
-              r.visible {|v| v[:percentil_75]<v[:high_whisker]}
+              r.visible {|v| v[:percentil_75] < v[:high_whisker]}
               r.bottom {|v| y_scale.scale(v[:high_whisker])}              
             end
             
              bp.rule do |r|
-              r.visible {|v| v[:percentil_75]<v[:high_whisker]}
+              r.visible {|v| v[:percentil_75] < v[:high_whisker]}
               r.bottom {|v| y_scale.scale(v[:percentil_75])}              
               r.left {|v| x_scale.range_band / 2.0}
-              r.height {|v| y_scale.scale(v[:high_whisker])-y_scale.scale(v[:percentil_75])}
+              r.height {|v| y_scale.scale(v[:high_whisker]) - y_scale.scale(v[:percentil_75])}
             end
-            
+            # Outliers
             bp.dot do |dot|
               dot.shape_size 4
               dot.data {|v| v[:outliers]}

data/lib/statsample/graph/histogram.rb

@@ -1,12 +1,26 @@
 require 'rubyvis'
 module Statsample
   module Graph
+    
+    # In statistics, a histogram is a graphical representation, showing a visual impression of the distribution of experimental data. It is an estimate of the probability distribution of a continuous variable and was first introduced by Karl Pearson [1]. A histogram consists of tabular frequencies, shown as adjacent rectangles, erected over discrete intervals (bins), with an area equal to the frequency of the observations in the interval. The height of a rectangle is also equal to the frequency density of the interval, i.e., the frequency divided by the width of the interval. The total area of the histogram is equal to the number of data.
+    # 
+    # == Usage
+    # === Svg output
+    #  a=[1,2,3,4].to_scale
+    #  puts Statsample::Graph::Histogram.new(a).to_svg
+    # === Using ReportBuilder
+    #  a=[1,2,3,4].to_scale
+    #  rb=ReportBuilder.new
+    #  rb.add(Statsample::Graph::Histogram.new(a))
+    #  rb.save_html('histogram.html')
+    
     class Histogram
       include Summarizable
+      # Histogram name
       attr_accessor :name
-      # Total width of Boxplot
+      # Total width
       attr_accessor :width
-      # Total height of Boxplot
+      # Total height
       attr_accessor :height
       # Top margin
       attr_accessor :margin_top
@@ -19,44 +33,91 @@ module Statsample
       attr_reader :hist
       # Could be an array of ranges or number of bins
       attr_accessor :bins
+      # Minimum value on x axis. Calculated automaticly from data if not set
+      attr_accessor :minimum_x
+      # Maximum value on x axis. Calculated automaticly from data if not set
+      attr_accessor :maximum_x
+      # Minimum value on y axis. Set to 0 if not set
+      attr_accessor :minimum_y
+      # Maximum value on y axis. Calculated automaticly from data if not set.
+      attr_accessor :maximum_y
+      # Add a line showing normal distribution
+      attr_accessor :line_normal_distribution
       # data could be a vector or a histogram
       def initialize(data,opts=Hash.new)
-        prov_name=data.respond_to? :name ? data.name : ""
+        prov_name=(data.respond_to?(:name)) ? data.name : ""
         opts_default={
           :name=>_("Histograma (%s)") % prov_name,
           :width=>400,
           :height=>300,
           :margin_top=>10,
           :margin_bottom=>20,
-          :margin_left=>20,
+          :margin_left=>30,
           :margin_right=>20,
-          :bins=>nil
+          :minimum_x=>nil,
+          :maximum_x=>nil,
+          :minimum_y=>nil,
+          :maximum_y=>nil,
+          :bins=>nil,
+          :line_normal_distribution=>false
         }
         @opts=opts_default.merge(opts)
         opts_default.keys.each {|k| send("#{k}=", @opts[k]) }
         @data=data
       end
-      def pre_vis
+      def pre_vis # :nodoc:
         if @data.is_a? Statsample::Histogram
           @hist=@data
+          @mean=@hist.estimated_mean
+          @sd=@hist.estimated_standard_deviation
         elsif @data.is_a? Statsample::Vector
+          @mean=@data.mean
+          @sd=@data.sd
           @bins||=Math::sqrt(@data.size).floor
           @hist=@data.histogram(@bins)
         end
       end
+      def rubyvis_normal_distribution(pan)
+        x_scale=@x_scale
+        y_scale=@y_scale
+        
+        wob = @hist.get_range(0)[1] - @hist.get_range(0)[0]
+        
+        nob = ((@maximum_x-@minimum_x) / wob.to_f).floor
+        sum=@hist.sum
+        
+        data=nob.times.map {|i|
+          l=@minimum_x+i*wob
+          r=@minimum_x+(i+1)*wob          
+          middle=(l+r) / 2.0
+          pi=Distribution::Normal.cdf((r-@mean) / @sd) - Distribution::Normal.cdf((l-@mean) / @sd)
+          {:x=>middle, :y=>pi*sum}
+        }
+        pan.line do |l|
+          l.data data
+          l.interpolate "cardinal"
+          l.stroke_style "black"
+          l.bottom {|d| y_scale[d[:y]]}
+          l.left {|d| x_scale[d[:x]]}
+        end
+        
+      end
       # Returns a Rubyvis panel with scatterplot
       def rubyvis_panel # :nodoc:
         pre_vis
         that=self
         
-        max_bin = @hist.max_val
+        @minimum_x||=@hist.min
+        @maximum_x||=@hist.max
+        @minimum_y||=0
+        @maximum_y||=@hist.max_val
         
         margin_hor=margin_left + margin_right
         margin_vert=margin_top  + margin_bottom
       
-        x_scale = pv.Scale.linear(@hist.min, @hist.max).range(0,width-margin_hor)
+        x_scale = pv.Scale.linear(@minimum_x, @maximum_x).range(0, width - margin_hor)
       
-        y_scale=Rubyvis::Scale.linear(0,max_bin).range(0, height-margin_vert)
+        y_scale=Rubyvis::Scale.linear(@minimum_y, @maximum_y).range(0, height - margin_vert)
         
         y_scale.nice
         max_range=@hist.max
@@ -67,6 +128,8 @@ module Statsample
            :value=>@hist.bin[i]
           }
         }
+        @x_scale=x_scale
+        @y_scale=y_scale
         # cache data
         vis=Rubyvis::Panel.new do |pan| 
           pan.width  width  - margin_hor
@@ -90,21 +153,22 @@ module Statsample
             left x_scale
             stroke_style "black"
             height 5
-            bottom -5
+            bottom(-5)
             label(:anchor=>'bottom') do
               text x_scale.tick_format
             end
           end
-          
+         
           pan.bar do |bar|
             bar.data(bins)
-            bar.left {|v| x_scale.scale(v[:low])}
-            bar.width {|v| x_scale.scale(v[:high]) - x_scale.scale(v[:low])}
+            bar.left {|v| x_scale[v[:low]]}
+            bar.width {|v| x_scale[v[:high]] - x_scale[v[:low]]}
             bar.bottom 0
-            bar.height {|v| y_scale.scale(v[:value])}
+            bar.height {|v| y_scale[v[:value]]}
             bar.stroke_style "black"
             bar.line_width 1
           end
+           rubyvis_normal_distribution(pan) if @line_normal_distribution
         end
       end
       # Returns SVG with scatterplot

data/lib/statsample/graph/scatterplot.rb

@@ -41,7 +41,25 @@ module Statsample
       
       attr_reader   :data
       attr_reader :v1,:v2
+      
+      # Array with assignation to groups of bars
+      # For example, for four vectors, 
+      #   boxplot.groups=[1,2,1,3]
+      # Assign same color to first and third element, and different to
+      # second and fourth
+      attr_accessor :groups
+
+      
       attr_reader :x_scale, :y_scale
+      # Minimum value on x axis. Calculated automaticly from data if not set
+      attr_accessor :minimum_x
+      # Maximum value on x axis. Calculated automaticly from data if not set
+      attr_accessor :maximum_x
+      # Minimum value on y axis. Set to 0 if not set
+      attr_accessor :minimum_y
+      # Maximum value on y axis. Calculated automaticly from data if not set.
+      attr_accessor :maximum_y
+
       # Create a new Scatterplot.
       # Params:
       # * v1: Vector on X axis
@@ -59,14 +77,18 @@ module Statsample
           :margin_top=>10,
           :margin_bottom=>20,
           :margin_left=>20,
-          :margin_right=>20
-          
+          :margin_right=>20,
+          :minimum_x=>nil,
+          :maximum_x=>nil,
+          :minimum_y=>nil,
+          :maximum_y=>nil,
+          :groups=>nil
         }
         @opts=opts_default.merge(opts)
         opts_default.keys.each {|k| send("#{k}=", @opts[k]) }
         @data=[]
         @v1.each_with_index {|d1,i|
-          @data.push({:x=>d1,:y=>@v2[i]})
+          @data.push({:x=>d1, :y=>@v2[i]})
         }
       end
       # Add a rule on median of X and Y axis
@@ -98,13 +120,24 @@ module Statsample
       def rubyvis_panel # :nodoc:
         that=self
         #p @v1.map {|v| v}
-        x=Rubyvis::Scale.linear(@v1.to_a).range(0,width)
-        y=Rubyvis::Scale.linear(@v2.to_a).range(0,height)
+        
+        @minimum_x||=@v1.min
+        @maximum_x||=@v1.max
+        @minimum_y||=@v2.min
+        @maximum_y||=@v2.max
+        
+        colors=Rubyvis::Colors.category10
+        
+        margin_hor=margin_left + margin_right
+        margin_vert=margin_top  + margin_bottom
+        
+        x=Rubyvis::Scale.linear(@minimum_x, @maximum_x).range(0, width - margin_hor)
+        y=Rubyvis::Scale.linear(@minimum_y, @maximum_y).range(0, height - margin_vert)
         @x_scale=x
         @y_scale=y
         vis=Rubyvis::Panel.new do |pan| 
-          pan.width  width  - (margin_left + margin_right)
-          pan.height height - (margin_top  + margin_bottom)
+          pan.width  width  - margin_hor
+          pan.height height - margin_vert
           pan.bottom margin_bottom
           pan.left   margin_left
           pan.right  margin_right
@@ -115,7 +148,7 @@ module Statsample
             bottom y
             stroke_style {|d| d!=0 ? "#eee" : "#000"}
             label(:anchor=>'left') do
-              visible {|d| d>0 and d<that.width}
+              visible {|d| d!=0 and  d < that.width}
               text y.tick_format
             end
           end
@@ -136,9 +169,26 @@ module Statsample
           pan.panel do
             data(that.data)
             dot do
-              left {|d| x.scale(d[:x])}
-              bottom {|d| y.scale(d[:y])}
-              stroke_style Rubyvis.color("red").alpha(that.dot_alpha)
+              left   {|d| x[d[:x]]}
+              bottom {|d| y[d[:y]]}
+              
+              fill_style {|v| 
+                alpha=(that.dot_alpha-0.3<=0) ? 0.1 : that.dot_alpha-0.3
+                if that.groups
+                  
+                  colors.scale(that.groups[index]).alpha(alpha)
+                else
+                  colors.scale(0).alpha(alpha)
+                end
+              }
+              
+              stroke_style {|v|
+                if that.groups
+                  colors.scale(that.groups[parent.index]).alpha(that.dot_alpha)
+                else
+                  colors.scale(0).alpha(that.dot_alpha)
+                end
+              }
               shape_radius 2
             end
           end

data/lib/statsample/histogram.rb

@@ -38,6 +38,7 @@ module Statsample
   # * http://www.gnu.org/software/gsl/manual/html_node/The-histogram-struct.html
   
     class Histogram
+      include Enumerable
       class << self
         # Alloc +n_bins+, using +range+ as ranges of bins
         def alloc(n_bins, range=nil, opts=Hash.new)
@@ -91,7 +92,7 @@ module Statsample
       end
       # 
       def increment(x, w=1)
-        if x.is_a? Array
+        if x.respond_to? :each
           x.each{|y| increment(y,w) }
         elsif x.is_a? Numeric
           (range.size-1).times do |i|
@@ -121,6 +122,41 @@ module Statsample
       def min_val
         @bin.min
       end
+      def each
+        bins.times.each do |i|
+          r=get_range(i)
+          arg={:i=>i, :low=>r[0],:high=>r[1], :middle=>(r[0]+r[1]) / 2.0,  :value=>@bin[i]}
+          yield arg
+        end
+      end
+      def estimated_variance
+        sum,n=0,0
+        mean=estimated_mean
+        each do |v|
+          sum+=v[:value]*(v[:middle]-mean)**2
+          n+=v[:value]
+        end
+        sum / (n-1)
+      end
+      def estimated_standard_deviation
+        Math::sqrt(estimated_variance)
+      end
+      def estimated_mean
+        sum,n=0,0
+        each do |v|
+          sum+= v[:value]* v[:middle]
+          n+=v[:value]
+        end
+        sum / n
+      end
+      alias :mean :estimated_mean
+      alias :sigma :estimated_standard_deviation
+      
+      def sum(start=nil,_end=nil)
+        start||=0
+        _end||=@n_bins-1
+        (start.._end).inject(0) {|ac,i| ac+@bin[i]}
+      end
       def report_building(generator)
         hg=Statsample::Graph::Histogram.new(self)
         generator.parse_element(hg)