Class: NumPlot::Dataset

Inherits:
Object
  • Object
show all
Defined in:
lib/numplot.rb

Overview

The dataset class plotted by Plotter.

You can construct a dataset object using one of the following three methods:

You need to choose from Dataset.columns and Dataset.rows according to the arrangement of your data.

The dataset object is registered to Plotter object using Plotter#add_dataset.

Parameters for a dataset

When you call Dataset.columns, Dataset.rows, or Dataset.histogram, you can give some parameters to determine the style of plotting. The parameters are as follows. If you know the details of these parameters, please read the gnuplot documents.

  • title: “TITLE” or notitle: true

  • with: “points”, “dots”, etc.

  • axis:

  • linestyle: a style number (ls: is also available)

  • linetype: a line type number

  • linewidth: a line width multiplier (default is 1.0), (lw: is also available)

  • linecolor: a color number or an RGB object. (lc: is also available)

  • pointtype: a point type number (pt: is also available)

  • pointsize: a point size multiplier (default is 1.0), (ls: is also available)

  • fill: a fill style (fs: is also available)

  • nohidden3d: true if you need to activate this option

  • nocontours: true if you need to activate this option

  • nosurface: true if you need to activate this option

Constant Summary

Class Method Summary (collapse)

Instance Method Summary (collapse)

Constructor Details

- (Dataset) initialize(data, opts = {})

A new instance of Dataset



496
497
498
499
 
# File 'lib/numplot.rb', line 496

def initialize(data, opts={})
  @data = data
  @opts = opts
end

Class Method Details

+ (Dataset) columns(data, opts = {})

Construct a dataset object form columns.

If your data are xs = [x0, x1, x2, … ] and ys = [y0, y1, y2, …], you should use this method as: Dataset.columns([xs, ys])

If your data is transposed, you need to use columns instead of this method.

You can give some parameters as a hash to set a plotting style. Please see NumPlot::Dataset.

Parameters:

  • data (Enumerable<Enumerable<Numeric>>)
  • opts ({Symbol => object}) (defaults to: {}) —

    options

Returns:



516
517
518
 
# File 'lib/numplot.rb', line 516

def self.columns(data, opts={})
  new(data.map(&:to_a).transpose, opts)
end

+ (Dataset) histogram(data, opts = {}) Also known as: histo

Construct a histogram dataset object.

You need to give some parameters to construct a histogram.

  • dim: the dimension of your data, now only 1-dim data is supported (default is 1)

  • binsize: the bin size (required, Numeric)

  • base: the coordinate of a center of a bin (default is 0.0). If binsize = 1.0 and base = 0.0, the bins are …, [-1.5, -0.5), [-0.5, 0.5), [0.5, 1.5), … . If binsize = 1.0 and base = 0.5, the bins are …, [-1.0, 0.0), [0.0, 1.0), … .

  • histogram_style: You can choose one of the following normalization scheme:

    • :count not normalized

    • :ratio normailized to the number of data points

    • :density normalized to (the number of data points)*(binsize). You choose this option if you want to show the probability density function from the data.

  • cumulative: construct a cumulative histogram if true. If the cumulative option is true, the meaning of histogram_style option is changed.

    • :count not normalized

    • :ratio, :density normalized by the number of data, this means that the maximum value of a cumulative histogram is normalized to 1.

You can also give some other parameters to determine a plotting style. See NumPlot::Dataset for details.

Examples:

require 'randomext' # install randomext gem
# 1000 random samples from the standard normal distribution.
rng = Random.new
data = Array.new(1000){ rng.standard_normal }
datasets = Dataset.histo(data, binsize: 0.1, histogram_style: :density,
                   title: "Random samples from the normal distribution")

Parameters:

  • data (Enumerable<Numeric>) —

    a sequence of numbers

  • opts ({Symbol => Object}) (defaults to: {}) —

    options

Returns:



599
600
601
602
603
604
605
606
607
608
 
# File 'lib/numplot.rb', line 599

def self.histogram(data, opts={})
  case opts.fetch(:dim, 1)
  when 1
    histo1d(data, opts)
  when 2
    raise NotImplementedError, "Histogram for 2D data is not implemented yet"
  else
    raise ArgumentError, "You can use only 1D or 2D histogram data"
  end
end

+ (Dataset) lines(data, opts = {})

Construct a dataset object from a sequence of sequences of rows

If your data is like [[[x00, y00] [x01, y01], … ], [[x10, y10], [x11, y11], …], ..] you should use this method.

You can give some parameters as a hash to set a plotting style. Please see NumPlot::Dataset for details.

Parameters:

  • data (Enumerable<Enumerable<Numeric>>)
  • opts ({Symbol => object}) (defaults to: {}) —

    options

Returns:



551
552
553
554
555
556
557
 
# File 'lib/numplot.rb', line 551

def self.lines(data, opts={})
  data = data.map{|lines|
    lines.to_a + [[]]
  }.flatten(1)
  p data
  new(data, opts)
end

+ (Dataset) rows(data, opts = {})

Construct a dataset object form rows.

If your data is [[x0, y0], [x1, y1], …], you should use this method as: Dataset.rows(data)

If your data is transposed, you need to use rows instead of this method.

You can give some parameters as a hash to set a plotting style. Please see NumPlot::Dataset for details.

Parameters:

  • data (Enumerable<Enumerable<Numeric>>)
  • opts ({Symbol => object}) (defaults to: {}) —

    options

Returns:



535
536
537
 
# File 'lib/numplot.rb', line 535

def self.rows(data, opts={})
  new(data.map(&:to_a), opts)
end