charty 0.2.7 → 0.2.11

data/lib/charty/plotters/abstract_plotter.rb

@@ -35,6 +35,8 @@ module Charty
       end
 
       def data=(data)
+        # TODO: Convert a Charty::Vector to a Charty::Table so that
+        # the Charty::Vector is handled as a wide form data
         @data = case data
                 when nil, Charty::Table
                   data
@@ -81,6 +83,24 @@ module Charty
                    end
       end
 
+      attr_reader :x_label
+
+      def x_label=(val)
+        @x_label = check_string(val, :x_label, allow_nil: true)
+      end
+
+      attr_reader :y_label
+
+      def y_label=(val)
+        @y_label = check_string(val, :y_label, allow_nil: true)
+      end
+
+      attr_reader :title
+
+      def title=(val)
+        @title = check_string(val, :title, allow_nil: true)
+      end
+
       private def substitute_options(options)
         options.each do |key, val|
           send("#{key}=", val)
@@ -138,6 +158,27 @@ module Charty
         end
       end
 
+      private def check_string(value, name, allow_nil: false)
+        case value
+        when Symbol
+          value.to_s
+        else
+          if allow_nil && value.nil?
+            nil
+          else
+            orig_value = value
+            value = String.try_convert(value)
+            if value.nil?
+              raise ArgumentError,
+                "`#{name}` must be convertible to String: %p" % orig_value,
+                caller
+            else
+              value
+            end
+          end
+        end
+      end
+
       private def variable_type(vector, boolean_type=:numeric)
         if vector.numeric?
           :numeric
@@ -181,15 +222,6 @@ module Charty
         data = processed ? processed_data : plot_data
         data = data.drop_na if drop_na
 
-        levels = var_levels.dup
-
-        ([:x, :y] & grouping_vars).each do |axis|
-          levels[axis] = plot_data[axis].categorical_order()
-          if processed
-            # TODO: perform inverse conversion of axis scaling here
-          end
-        end
-
         if not grouping_vars.empty?
           grouped = data.group_by(grouping_vars, sort: false)
           grouped.each_group do |group_key, group_data|

data/lib/charty/plotters/bar_plotter.rb

@@ -42,6 +42,12 @@ module Charty
         @cap_size = check_number(cap_size, :cap_size, allow_nil: true)
       end
 
+      attr_reader :log
+
+      def log=(val)
+        @log = check_boolean(val, :log)
+      end
+
       private def render_plot(backend, **)
         draw_bars(backend)
         annotate_axes(backend)
@@ -81,6 +87,39 @@ module Charty
         end
       end
 
+      private def annotate_axes(backend)
+        super
+
+        if self.log
+          min_value, max_value = @estimations.minmax
+          if @plot_colors
+            unless @conf_int.empty?
+              min_value = [min_value, @conf_int[0]].min
+              max_value = [max_value, @conf_int[1]].max
+            end
+          else
+            ci_min = Util.filter_map(@conf_int) { |ci| ci[0] unless ci.empty? }
+            ci_max = Util.filter_map(@conf_int) { |ci| ci[1] unless ci.empty? }
+            min_value = [min_value, ci_min.min].min unless ci_min.empty?
+            max_value = [max_value, ci_max.max].max unless ci_max.empty?
+          end
+          if min_value > 1
+            min_value = 0
+          else
+            min_value = Math.log10(min_value).floor
+          end
+          max_value = Math.log10(max_value).ceil
+          case self.orient
+          when :v
+            backend.set_yscale(:log)
+            backend.set_ylim(min_value, max_value)
+          else
+            backend.set_xscale(:log)
+            backend.set_xlim(min_value, max_value)
+          end
+        end
+      end
+
       private def setup_estimations
         if @color_names.nil?
           setup_estimations_with_single_color_group

data/lib/charty/plotters/categorical_plotter.rb

@@ -134,6 +134,7 @@ module Charty
         order = @order # TODO: supply order via parameter
         unless order
           order = @data.column_names.select do |cn|
+            # TODO: Use Charty::Vector#numeric?
             @data[cn].all? {|x| Float(x, exception: false) }
           end
         end
@@ -230,6 +231,7 @@ module Charty
           end
           return :h
         end
+
         case orient
         when :v
           if require_numeric && y_type != :numeric
@@ -263,7 +265,9 @@ module Charty
       private def group_long_form(vals, groups, group_order)
         grouped_vals = vals.group_by(groups)
 
-        plot_data = group_order.map {|g| grouped_vals[g] || [] }
+        plot_data = group_order.map do |g|
+          grouped_vals[g] || Charty::Vector.new([])
+        end
 
         if vals.respond_to?(:name)
           value_label = vals.name
@@ -347,11 +351,15 @@ module Charty
       end
 
       private def annotate_axes(backend)
+        backend.set_title(self.title) if self.title
+
         if orient == :v
           xlabel, ylabel = @group_label, @value_label
         else
           xlabel, ylabel = @value_label, @group_label
         end
+        xlabel = self.x_label if self.x_label
+        ylabel = self.y_label if self.y_label
         backend.set_xlabel(xlabel) unless xlabel.nil?
         backend.set_ylabel(ylabel) unless ylabel.nil?
 

data/lib/charty/plotters/distribution_plotter.rb

@@ -3,7 +3,14 @@ module Charty
     class DistributionPlotter < AbstractPlotter
       def flat_structure
         {
-          x: :values
+          x: :@values
+        }
+      end
+
+      def wide_structure
+        {
+          x: :@values,
+          color: :@columns
         }
       end
 
@@ -14,6 +21,12 @@ module Charty
         setup_variables
       end
 
+      attr_reader :weights
+
+      def weights=(val)
+        @weights = check_dimension(val, :weights)
+      end
+
       attr_reader :variables
 
       attr_reader :color_norm
@@ -65,7 +78,6 @@ module Charty
           return
         end
 
-        # TODO: detect flat data
         flat = data.is_a?(Charty::Vector)
         if flat
           @plot_data = {}
@@ -73,10 +85,10 @@ module Charty
 
           [:x, :y].each do |var|
             case self.flat_structure[var]
-            when :index
+            when :@index
               @plot_data[var] = data.index.to_a
               @variables[var] = data.index.name
-            when :values
+            when :@values
               @plot_data[var] = data.to_a
               @variables[var] = data.name
             end
@@ -84,8 +96,32 @@ module Charty
 
           @plot_data = Charty::Table.new(@plot_data)
         else
-          raise NotImplementedError,
-                "wide-form input is not supported"
+          numeric_columns = @data.column_names.select do |cn|
+            @data[cn].numeric?
+          end
+          wide_data = @data[numeric_columns]
+
+          melt_params = {var_name: :@columns, value_name: :@values }
+          if self.wide_structure.include?(:index)
+            melt_params[:id_vars] = :@index
+          end
+
+          @plot_data = wide_data.melt(**melt_params)
+          @variables = {}
+          self.wide_structure.each do |var, attr|
+            @plot_data[var] = @plot_data[attr]
+
+            @variables[var] = case attr
+                              when :@columns
+                                wide_data.columns.name
+                              when :@index
+                                wide_data.index.name
+                              else
+                                nil
+                              end
+          end
+
+          @plot_data = @plot_data[self.wide_structure.keys]
         end
       end
 
@@ -103,10 +139,11 @@ module Charty
           x: self.x,
           y: self.y,
           color: self.color,
+          weights: self.weights
         }.each do |key, val|
           next if val.nil?
 
-          if data.column_names.include?(val)
+          if data.column?(val)
             plot_data[key] = data[val]
             variables[key] = val
           else

data/lib/charty/plotters/histogram_plotter.rb

@@ -12,16 +12,6 @@ module Charty
         ([:x, :y] & self.variables.keys)[0]
       end
 
-      attr_reader :weights
-
-      def weights=(val)
-        @weights = check_weights(val)
-      end
-
-      private def check_weights(val)
-        raise NotImplementedError, "weights is not supported yet"
-      end
-
       attr_reader :stat
 
       def stat=(val)
@@ -65,10 +55,44 @@ module Charty
       end
 
       # TODO: bin_width
-      # TODO: bin_range
+
+      attr_reader :bin_range
+
+      def bin_range=(val)
+        @bin_range = check_bin_range(val)
+      end
+
+      private def check_bin_range(val)
+        case val
+        when nil, Range
+          return val
+        when Array
+          if val.length == 2
+            val.each_with_index do |v, i|
+              check_number(v, "bin_range[#{i}]")
+            end
+            return val
+          else
+            amount = val.length < 2 ? "few" : "many"
+            raise ArgumentError,
+                  "Too #{amount} items in `bin_range` array (%p for 2)" % val.length
+          end
+        else
+          raise ArgumentError,
+                "Invalid value for `bin_range` " +
+                "(%p for a range or a pair of numbers)" % val
+        end
+      end
+
       # TODO: discrete
       # TODO: cumulative
-      # TODO: common_bins
+
+      attr_reader :common_bins
+
+      def common_bins=(val)
+        @common_bins = check_boolean(val, :common_bins)
+      end
+
       # TODO: common_norm
 
       attr_reader :multiple
@@ -127,24 +151,52 @@ module Charty
       private def draw_univariate_histogram(backend)
         map_color(palette: palette, order: color_order, norm: color_norm)
 
+        key_color = self.key_color
+        if key_color.nil? && !self.variables.key?(:color)
+          palette = case self.palette
+                    when Palette
+                      self.palette
+                    when nil
+                      Palette.default
+                    else
+                      Palette[self.palette]
+                    end
+          key_color = palette[0]
+        end
+
         # TODO: calculate histogram here and use bar plot to visualize
         data_variable = self.univariate_variable
 
-        histograms = {}
-        each_subset([:color], processed: true) do |sub_vars, sub_data|
-          key = sub_vars.to_a
-          observations = sub_data[data_variable].drop_na.to_a
-          hist = Statistics.histogram(observations)
-          histograms[key] = hist
-        end
-
-        bin_start, bin_end, bin_size = nil
-        histograms.each do |_, hist|
-          s, e = hist.edge.minmax
-          z = (e - s).to_f / (hist.edge.length - 1)
-          bin_start = [bin_start, s].compact.min
-          bin_end   = [bin_end, e].compact.max
-          bin_size  = [bin_size, z].compact.min
+        if common_bins
+          all_data = processed_data.drop_na
+          all_observations = all_data[data_variable].to_a
+
+          bins = self.bins
+          bins = 10 if self.variables.key?(:color) && bins == :auto
+
+          case bins
+          when Integer
+            case bin_range
+            when Range
+              start = bin_range.begin
+              stop  = bin_range.end
+            when Array
+              start, stop = bin_range.minmax
+            end
+            data_range = all_observations.minmax
+            start ||= data_range[0]
+            stop ||= data_range[1]
+            if start == stop
+              start -= 0.5
+              stop += 0.5
+            end
+            common_bin_edges = Linspace.new(start .. stop, bins + 1).map(&:to_f)
+          else
+            params = {}
+            params[:weights] = all_data[:weights].to_a if all_data.column?(:weights)
+            h = Statistics.histogram(all_observations, bins, **params)
+            common_bin_edges = h.edges
+          end
         end
 
         if self.variables.key?(:color)
@@ -154,27 +206,37 @@ module Charty
         end
 
         each_subset([:color], processed: true) do |sub_vars, sub_data|
-          name = sub_vars[:color]
           observations = sub_data[data_variable].drop_na.to_a
+          params = {}
+          params[:weights] = sub_data[:weights].to_a if sub_data.column?(:weights)
+          params[:edges] = common_bin_edges if common_bin_edges
+          hist = Statistics.histogram(observations, bins, **params)
 
-          backend.univariate_histogram(observations, name, data_variable, stat,
-                                       bin_start, bin_end, bin_size, alpha,
-                                       name, @color_mapper)
+          name = sub_vars[:color]
+          backend.univariate_histogram(hist, name, data_variable, stat,
+                                       alpha, name, key_color, @color_mapper,
+                                       multiple, :bars, true, 1r)
         end
       end
 
       private def annotate_axes(backend)
+        backend.set_title(self.title) if self.title
+
         if univariate?
-          xlabel = self.variables[:x]
-          ylabel = self.variables[:y]
+          xlabel = self.x_label || self.variables[:x]
+          ylabel = self.y_label || self.variables[:y]
           case self.univariate_variable
           when :x
-            ylabel = self.stat.to_s.capitalize
+            ylabel ||= self.stat.to_s.capitalize
           else
-            xlabel = self.stat.to_s.capitalize
+            xlabel ||= self.stat.to_s.capitalize
           end
           backend.set_ylabel(ylabel) if ylabel
           backend.set_xlabel(xlabel) if xlabel
+
+          if self.variables.key?(:color)
+            backend.legend(loc: :best, title: self.variables[:color])
+          end
         end
       end
     end

data/lib/charty/plotters/line_plotter.rb

@@ -122,7 +122,7 @@ module Charty
 
       include RandomSupport
 
-      attr_reader :sort, :err_style, :err_kws, :error_bar, :x_scale, :y_scale
+      attr_reader :sort, :err_style, :err_kws, :error_bar
 
       def sort=(val)
         @sort = check_boolean(val, :sort)
@@ -211,17 +211,21 @@ module Charty
         [method, level]
       end
 
+      attr_reader :x_scale
+
       def x_scale=(val)
         @x_scale = check_axis_scale(val, :x)
       end
 
+      attr_reader :y_scale
+
       def y_scale=(val)
         @y_scale = check_axis_scale(val, :y)
       end
 
       private def check_axis_scale(val, axis)
         case val
-        when :linear, "linear", :log10, "log10"
+        when :linear, "linear", :log, "log"
           val.to_sym
         else
           raise ArgumentError,
@@ -252,6 +256,15 @@ module Charty
             sub_data = sub_data.sort_values(sort_cols)
           end
 
+          # Perform axis scaling
+          if x_scale != :linear
+            sub_data[:x] = sub_data[:x].scale(x_scale)
+          end
+          if y_scale != :linear
+            sub_data[:y] = sub_data[:y].scale(x_scale)
+          end
+
+          # Perform estimation and error calculation
           unless estimator.nil?
             if self.variables.include?(:units)
               raise "`estimator` is must be nil when specifying `units`"
@@ -261,7 +274,22 @@ module Charty
             sub_data = grouped.apply(agg_var, &aggregator.method(:aggregate)).reset_index
           end
 
-          # TODO: perform inverse conversion of axis scaling before plot
+          # Perform axis inverse scaling
+          if x_scale != :linear
+            sub_data.column_names.each do |cn|
+              if cn.start_with?("x")
+                sub_data[cn] = sub_data[cn].scale_inverse(x_scale)
+              end
+            end
+          end
+
+          if y_scale != :linear
+            sub_data.column_names.each do |cn|
+              if cn.start_with?("y")
+                sub_data[cn] = sub_data[cn].scale_inverse(x_scale)
+              end
+            end
+          end
 
           unit_grouping = if self.variables.include?(:units)
                             sub_data.group_by(:units).each_group
@@ -290,10 +318,15 @@ module Charty
       end
 
       private def annotate_axes(backend)
-        xlabel = self.variables[:x]
-        ylabel = self.variables[:y]
+        backend.set_title(self.title) if self.title
+
+        xlabel = self.x_label || self.variables[:x]
+        ylabel = self.y_label || self.variables[:y]
         backend.set_xlabel(xlabel) unless xlabel.nil?
         backend.set_ylabel(ylabel) unless ylabel.nil?
+
+        backend.set_xscale(x_scale)
+        backend.set_yscale(y_scale)
       end
     end
   end