View Source Tucan (tucan v0.2.0)
A high level API interface for creating plots on top of VegaLite.
Tucan is an Elixir plotting library built on top of VegaLite,
designed to simplify the creation of interactive and visually stunning
plots. With Tucan, you can effortlessly generate a wide range of plots,
from simple bar charts to complex composite plots, all while enjoying the
power and flexibility of a clean composable functional API.
Tucan offers a simple API for creating most common plot types similarly
to the widely used python packages matplotlib and seaborn without
requiring the end user to be familiar with the Vega Lite grammar.
Features
- Versatile Plot Types - Tucan provides an array of plot types, including bar charts, line plots, scatter plots, histograms, and more, allowing you to effectively represent diverse data sets.
- Clean and consistent API - A clean and consistent plotting API similar
to
matplotliborseabornis provided. You should be able to create most common plots with a single function call and minimal configuration. - Grouping and Faceting - Enhance your visualizations with grouping and faceting features, enabling you to examine patterns and trends within subgroups of your data.
- Customization - Customize your plots with ease using Tucan's utilities for adjusting plot dimensions, titles, and themes.
- Thin wrapper on top of VegaLite - All
VegaLitefunctions can be used seamlessly withTucanfor advanced customizations if needed. - Low level API - A low level API with helper functions allow you to modify
any part of a
VegaLitespecification.
Basic usage
All supported plots expect as first argument some data, a VegaLite specification
or a binary which is considered a url to some data. Additionally you can use
one of the available Tucan.Datasets.
Tucan.scatter(:iris, "petal_width", "petal_length")You can apply semantic grouping by a third variable by modifying the color, the shape or the size of the points:
Tucan.scatter(:iris, "petal_width", "petal_length", color_by: "species", shape_by: "species")Alternatively you could use the helper grouping functions:
Tucan.scatter(:iris, "petal_width", "petal_length")
|> Tucan.color_by("species")
|> Tucan.shape_by("species")Use the functional API carefully
For some plot types where transformations are applied on the input data it is recommended to use the options instead of the functional API, since in the first case any required grouping will also be applied to the transformations.
Composite plots
Tucan also provides some helper functions for generating composite plots.
pairplot/3 can be used to plot pairwise relationships across a dataset.
fields = ["Beak Length (mm)", "Beak Depth (mm)", "Body Mass (g)"]
Tucan.pairplot(:penguins, fields, diagonal: :density)Customization & Themes
Various methods and helper modules allow you to easily modify the style of a plot.
Tucan.bubble(:gapminder, "income", "health", "population",
color_by: "region",
width: 400,
tooltip: :data
)
|> Tucan.Axes.set_x_title("Gdp per Capita")
|> Tucan.Axes.set_y_title("Life expectancy")
|> Tucan.Scale.set_x_scale(:log)
|> Tucan.Grid.set_color(:x, "red")Additionally set_theme/2 allows you to set one of the supported Tucan.Themes.
Tucan.density_heatmap(:penguins, "Beak Length (mm)", "Beak Depth (mm)")
|> Tucan.set_theme(:latimes)Encoding channels options
All Tucan plots are building a VegaLite specification based on some sane
default parameters. Notice that only a tiny subset of vega-lite configuration
options are exported in Tucan's public API. This is more than enough in most
cases. Additionally, an optional configuration option is added for every
encoding channel that is used, that allows you to add any vega-lite option
or change the default options set by Tucan.
For example:
Tucan.bar(:weather, "date", "date",
color_by: "weather",
tooltip: true,
x: [type: :ordinal, time_unit: :month],
y: [aggregate: :count]
)Summary
Plots
Returns the specification of an area plot.
Returns the specification of a bar chart.
Returns the specification of a box plot.
Returns the specification of a bubble plot.
Plot the counts of observations for a categorical variable.
Plot the distribution of a numeric variable.
Draws a density heatmap.
Draw a donut chart.
Returns the specification of a heatmap.
Plots a histogram.
Draw a line plot between x and y
Draws a pie chart.
Returns the specification of a punch card plot.
Returns the specification of a scatter plot with possibility of several semantic groupings.
Returns the specification of a step chart.
Returns the specification of a streamgraph.
Draws a strip plot (categorical scatterplot).
Grouping
Adds a color encoding for the given field.
Apply facetting on the input plot vl by the given field.
Adds a fill encoding for the given field.
Adds a shape encoding for the given field.
Adds a size encoding for the given field.
Adds a stroke_dash encoding for the given field.
Utilities
Concatenates the given plots.
Flips the axes of the provided chart.
Concatenates horizontally the given plots.
Adds a horizontal line at the given h position.
Creates a layered plot.
Creates if needed a VegaLite plot and adds data to it.
Adds a vertical or horizontal ruler at the given position.
Concatenates vertically the given plots.
Adds a vertical line at the given x position.
Styling
Sets the height of the plot (in pixels).
Sets the plot size.
Sets the plot's theme.
Sets the title of the plot.
Sets the width of the plot (in pixels).
Types
@type field() :: binary()
@type plotdata() :: binary() | Table.Reader.t() | Tucan.Datasets.t() | VegaLite.t()
Plots
@spec area(plotdata :: plotdata(), x :: field(), y :: field(), opts :: keyword()) :: VegaLite.t()
Returns the specification of an area plot.
Options
:interpolate(binary/0) - The line interpolation method to use for line and area marks. One of the following:"linear"- piecewise linear segments, as in a poly-line."linear-closed"- close the linear segments to form a polygon."step"- alternate between horizontal and vertical segments, as in a step function."step-before"- alternate between vertical and horizontal segments, as in a step function."step-after"- alternate between horizontal and vertical segments, as in a step function."basis"- a B-spline, with control point duplication on the ends."basis-open"- an open B-spline; may not intersect the start or end."basis-closed"- a closed B-spline, as in a loop."cardinal"- a Cardinal spline, with control point duplication on the ends."cardinal-open"- an open Cardinal spline; may not intersect the start or end, but will intersect other control points."cardinal-closed"- a closed Cardinal spline, as in a loop."bundle"- equivalent to basis, except the tension parameter is used to straighten the spline."monotone"- cubic interpolation that preserves monotonicity in y.
:line(boolean/0) - Whether the line will be included in the chart The default value isfalse.:mode- The stacking mode, applied only if:color_byis set. Can be one of the following::stacked- the default one, areas are stacked:normalize- the stacked charts are normalized:streamgraph- the chart is displaced around a central axis:no_stack- no stacking is applied The default value is:stacked.
:points(boolean/0) - Whether points will be included in the chart. The default value isfalse.:tension(number/0) - Depending on the interpolation type, sets the tension parameter
Data Grouping Options
:color_by(String.t/0) - If set a data field that will be used for coloring the data. It is considered:nominalby default.
Styling Options
:clip(boolean/0) - Whether a mark will be clipped to the enclosing group’s width and height.:fill_opacity(number/0) - The fill opacity of the plotted elements. The default value is1.:height(integer/0) - Height of the image:title(String.t/0) - The title of the graph:width(integer/0) - Width of the image
Encodings Custom Options
All Tucan plots are building a VegaLite specification based on some sane
default parameters. Through these encodings options you are free to set any
vega-lite supported option to any encoding channel of the plot.
Notice that if set they will be merged with any option set by the plot. Since they have a higher precedence they may override the default settings and affect the generated plot.
You can set an arbitrary keyword list. Notice that the contents are not validated.
:color(keyword/0) - Extra vega lite options for the:colorencoding. The default value is[].:x(keyword/0) - Extra vega lite options for the:xencoding. The default value is[].:y(keyword/0) - Extra vega lite options for the:yencoding. The default value is[].
Interactivity Options
:tooltip(boolean() | :data | :encoding) - The tooltip text string to show upon mouse hover or an object defining which fields should the tooltip be derived from. Can be one of the following::encoding- all fields from encoding are used:data- all fields of the highlighted data point are usedtrue- same as:encodingfalse,nil- no tooltip is used
Examples
A simple area chart of Google stock price over time. Notice how we change the
x axis type from the default (:quantitative) to :temporal using the generic
:x channel configuration option:
Tucan.area(:stocks, "date", "price", x: [type: :temporal])
|> VegaLite.transform(filter: "datum.symbol==='GOOG'")You can overlay the points and/or the line:
Tucan.area(:stocks, "date", "price", x: [type: :temporal], points: true, line: true)
|> VegaLite.transform(filter: "datum.symbol==='GOOG'")If you add the :color_by property then the area charts are stacked by default. Below
you can see how the generic encoding options can be used in order to modify any part
of the underlying VegaLite specification:
Tucan.area(:unemployment, "date", "count",
color_by: "series",
x: [type: :temporal, time_unit: :yearmonth, axis: [format: "%Y"]],
y: [aggregate: :sum],
color: [scale: [scheme: "category20b"]],
width: 300,
height: 200
)You could change the mode to :normalize or :streamgraph:
left =
Tucan.area(:unemployment, "date", "count",
color_by: "series",
mode: :normalize,
x: [type: :temporal, time_unit: :yearmonth, axis: [format: "%Y"]],
y: [aggregate: :sum]
)
|> Tucan.set_title("normalize")
right =
Tucan.area(:unemployment, "date", "count",
color_by: "series",
mode: :streamgraph,
x: [type: :temporal, time_unit: :yearmonth, axis: [format: "%Y"]],
y: [aggregate: :sum]
)
|> Tucan.set_title("streamgraph")
Tucan.hconcat([left, right])Or you could disable the stacking at all:
Tucan.area(:stocks, "date", "price",
color_by: "symbol",
mode: :no_stack,
x: [type: :temporal],
width: 400,
fill_opacity: 0.4
)
|> Tucan.Scale.set_y_scale(:log)@spec bar( plotdata :: plotdata(), field :: binary(), value :: binary(), opts :: keyword() ) :: VegaLite.t()
Returns the specification of a bar chart.
A bar chart is consisted by a categorical field and a numerical value field that
defines the height of the bars. You can create a grouped bar chart by setting
the :color_by option.
Additionally you should specify the aggregate for the y values, if your dataset contains
more than one values per category.
Options
:mode- The stacking mode, applied only if:color_byis set. Can be one of the following::stacked- the default one, bars are stacked:normalize- the bars are stacked are normalized:grouped- no stacking is applied, a separate bar for each category The default value is:stacked.
Data Grouping Options
:color_by(String.t/0) - If set a data field that will be used for coloring the data. It is considered:nominalby default.
Styling Options
:clip(boolean/0) - Whether a mark will be clipped to the enclosing group’s width and height.:fill_opacity(number/0) - The fill opacity of the plotted elements. The default value is1.:height(integer/0) - Height of the image:orient(atom/0) - The plot's orientation, can be either:horizontalor:vertical. The default value is:horizontal.:title(String.t/0) - The title of the graph:width(integer/0) - Width of the image
Encodings Custom Options
All Tucan plots are building a VegaLite specification based on some sane
default parameters. Through these encodings options you are free to set any
vega-lite supported option to any encoding channel of the plot.
Notice that if set they will be merged with any option set by the plot. Since they have a higher precedence they may override the default settings and affect the generated plot.
You can set an arbitrary keyword list. Notice that the contents are not validated.
:color(keyword/0) - Extra vega lite options for the:colorencoding. The default value is[].:x(keyword/0) - Extra vega lite options for the:xencoding. The default value is[].:x_offset(keyword/0) - Extra vega lite options for the:x_offsetencoding. The default value is[].:y(keyword/0) - Extra vega lite options for the:yencoding. The default value is[].
Interactivity Options
:tooltip(boolean() | :data | :encoding) - The tooltip text string to show upon mouse hover or an object defining which fields should the tooltip be derived from. Can be one of the following::encoding- all fields from encoding are used:data- all fields of the highlighted data point are usedtrue- same as:encodingfalse,nil- no tooltip is used
Examples
A simple bar chart:
data = [
%{"a" => "A", "b" => 28},
%{"a" => "B", "b" => 55},
%{"a" => "C", "b" => 43},
%{"a" => "D", "b" => 91},
%{"a" => "E", "b" => 81},
%{"a" => "F", "b" => 53},
%{"a" => "G", "b" => 19},
%{"a" => "H", "b" => 87},
%{"a" => "I", "b" => 52}
]
Tucan.bar(data, "a", "b")You can set a color_by option that will create a stacked bar chart:
Tucan.bar(:weather, "date", "date",
color_by: "weather",
tooltip: true,
x: [type: :ordinal, time_unit: :month],
y: [aggregate: :count]
)If you set the mode option to :grouped you will instead have a different bar
per group, you can also change the orientation by setting the :orient flag.
Similarly you can set the mode to :normalize in order to have normalized
stacked bars.
data = [
%{"category" => "A", "group" => "x", "value" => 0.1},
%{"category" => "A", "group" => "y", "value" => 0.6},
%{"category" => "A", "group" => "z", "value" => 0.9},
%{"category" => "B", "group" => "x", "value" => 0.7},
%{"category" => "B", "group" => "y", "value" => 0.2},
%{"category" => "B", "group" => "z", "value" => 1.1},
%{"category" => "C", "group" => "x", "value" => 0.6},
%{"category" => "C", "group" => "y", "value" => 0.1},
%{"category" => "C", "group" => "z", "value" => 0.2}
]
grouped =
Tucan.bar(
data,
"category",
"value",
color_by: "group",
mode: :grouped,
orient: :vertical
)
normalized =
Tucan.bar(
data,
"category",
"value",
color_by: "group",
mode: :normalize
)
Tucan.hconcat([grouped, normalized])@spec boxplot(plotdata :: plotdata(), field :: binary(), opts :: keyword()) :: VegaLite.t()
Returns the specification of a box plot.
By default a one dimensional box plot of the :field - which must be a numerical variable - is
generated. You can add a second dimension across a categorical variable by either setting the
:group or :color_by options.
By default a Tukey box plot will be generated. In the Tukey box plot the whisker spans from
the smallest data to the largest data within the range [Q1 - k * IQR, Q3 + k * IQR] where
Q1and Q3 are the first and third quartiles while IQR is the interquartile range
(Q3-Q1). You can specify if needed the constant k which defaults to 1.5.
Additionally you can set the mode to :min_max where the lower and upper whiskers are
defined as the min and max respectively. No points will be considered as outliers for this
type of box plots. In this case the k value is ignored.
What is a box plot
A box plot (box and whisker plot) displays the five-number summary of a set of data. The five-number summary is the minimum, first quartile, median, third quartile, and maximum. In a box plot, we draw a box from the first quartile to the third quartile. A vertical line goes through the box at the median.
Options
:k(float/0) - The constant used for calculating the extent of the whiskers in a Tukey boxplot. Applicable only if:modeis set to:tukey. The default value is1.5.:mode- The type of the box plot. Either a Tukey box plot will be created or a min-max plot. The default value is:tukey.
Data Grouping Options
:color_by(String.t/0) - If set a data field that will be used for coloring the data. It is considered:nominalby default.:group_by(String.t/0) - A field to be used for grouping the boxplot. It is used for adding a second dimension to the plot. If not set the plot will be one dimensional. Notice that a grouping is automatically applied if the:color_byoption is set.
Styling Options
:clip(boolean/0) - Whether a mark will be clipped to the enclosing group’s width and height.:fill_opacity(number/0) - The fill opacity of the plotted elements. The default value is1.:height(integer/0) - Height of the image:orient(atom/0) - The plot's orientation, can be either:horizontalor:vertical. The default value is:horizontal.:title(String.t/0) - The title of the graph:width(integer/0) - Width of the image
Encodings Custom Options
All Tucan plots are building a VegaLite specification based on some sane
default parameters. Through these encodings options you are free to set any
vega-lite supported option to any encoding channel of the plot.
Notice that if set they will be merged with any option set by the plot. Since they have a higher precedence they may override the default settings and affect the generated plot.
You can set an arbitrary keyword list. Notice that the contents are not validated.
:color(keyword/0) - Extra vega lite options for the:colorencoding. The default value is[].:x(keyword/0) - Extra vega lite options for the:xencoding. The default value is[].:y(keyword/0) - Extra vega lite options for the:yencoding. The default value is[].
Interactivity Options
:tooltip(boolean() | :data | :encoding) - The tooltip text string to show upon mouse hover or an object defining which fields should the tooltip be derived from. Can be one of the following::encoding- all fields from encoding are used:data- all fields of the highlighted data point are usedtrue- same as:encodingfalse,nil- no tooltip is used
Examples
A one dimensional Tukey boxplot:
Tucan.boxplot(:penguins, "Body Mass (g)")You can set :group or :color_by in order to set a second dimension:
Tucan.boxplot(:penguins, "Body Mass (g)", color_by: "Species")You can set the mode to :min_max in order to extend the whiskers to the min and max values:
Tucan.boxplot(:penguins, "Body Mass (g)", color_by: "Species", mode: :min_max)By setting the :orient to :vertical you can change the default horizontal orientation:
Tucan.boxplot(:penguins, "Body Mass (g)", color_by: "Species", orient: :vertical)@spec bubble( plotdata :: plotdata(), x :: field(), y :: field(), size :: field(), opts :: keyword() ) :: VegaLite.t()
Returns the specification of a bubble plot.
A bubble plot is a scatter plot with a third parameter defining the size of the dots.
All x, y and size must be numerical data fields.
See also scatter/4.
Options
Data Grouping Options
:color_by(String.t/0) - If set a data field that will be used for coloring the data. It is considered:nominalby default.
Styling Options
:clip(boolean/0) - Whether a mark will be clipped to the enclosing group’s width and height.:fill_opacity(number/0) - The fill opacity of the plotted elements. The default value is1.:height(integer/0) - Height of the image:title(String.t/0) - The title of the graph:width(integer/0) - Width of the image
Encodings Custom Options
All Tucan plots are building a VegaLite specification based on some sane
default parameters. Through these encodings options you are free to set any
vega-lite supported option to any encoding channel of the plot.
Notice that if set they will be merged with any option set by the plot. Since they have a higher precedence they may override the default settings and affect the generated plot.
You can set an arbitrary keyword list. Notice that the contents are not validated.
:color(keyword/0) - Extra vega lite options for the:colorencoding. The default value is[].:size(keyword/0) - Extra vega lite options for the:sizeencoding. The default value is[].:x(keyword/0) - Extra vega lite options for the:xencoding. The default value is[].:y(keyword/0) - Extra vega lite options for the:yencoding. The default value is[].
Interactivity Options
:tooltip(boolean() | :data | :encoding) - The tooltip text string to show upon mouse hover or an object defining which fields should the tooltip be derived from. Can be one of the following::encoding- all fields from encoding are used:data- all fields of the highlighted data point are usedtrue- same as:encodingfalse,nil- no tooltip is used
Examples
Tucan.bubble(:gapminder, "income", "health", "population", width: 400)
|> Tucan.Axes.set_x_title("Gdp per Capita")
|> Tucan.Axes.set_y_title("Life expectancy")You could use a fourth variable to color the graph. As always you can set the tooltip in
order to make the plot interactive:
Tucan.bubble(:gapminder, "income", "health", "population",
color_by: "region",
width: 400,
tooltip: :data
)
|> Tucan.Axes.set_x_title("Gdp per Capita")
|> Tucan.Axes.set_y_title("Life expectancy")It makes more sense to use a log scale for the x axis:
Tucan.bubble(:gapminder, "income", "health", "population",
color_by: "region",
width: 400,
tooltip: :data
)
|> Tucan.Axes.set_x_title("Gdp per Capita")
|> Tucan.Axes.set_y_title("Life expectancy")
|> Tucan.Scale.set_x_scale(:log)@spec countplot(plotdata :: plotdata(), field :: binary(), opts :: keyword()) :: VegaLite.t()
Plot the counts of observations for a categorical variable.
Takes a categorical field as input and generates a count plot
visualization. By default the counts are plotted on the y-axis
and the categorical field across the x-axis.
This is similar to histogram/3 but specifically for a categorical
variable.
This is a simple wrapper around bar/4 where by default the count of
observations is mapped to the y variable.
What is a countplot?
A countplot is a type of bar chart used in data visualization to display the frequency of occurrences of categorical data. It is particularly useful for visualizing the distribution and frequency of different categories within a dataset.
In a countplot, each unique category is represented by a bar, and the height of the bar corresponds to the number of occurrences of that category in the data.
Options
See bar/4
Examples
We will use the :titanic dataset on the following examples. We can
plot the number of passengers by ticket class:
Tucan.countplot(:titanic, "Pclass")You can make the bars horizontal by setting the :orient option:
Tucan.countplot(:titanic, "Pclass", orient: :vertical)You can set :color_by to group it by a second variable:
Tucan.countplot(:titanic, "Pclass", color_by: "Survived")By default the bars are stacked. You can unstack them by setting the
:mode to :grouped
Tucan.countplot(:titanic, "Pclass", color_by: "Survived", mode: :grouped)@spec density(plotdata :: plotdata(), field :: binary(), opts :: keyword()) :: VegaLite.t()
Plot the distribution of a numeric variable.
Density plots allow you to visualize the distribution of a numeric variable for one
or several groups. If you want to draw the density for several groups you need to
specify the :color_by option which is assumed to be a categorical variable.
Avoid calling
color_by/3with a density plotSince the grouping variable must also be used for properly calculating the density transformation you should avoid calling the
color_by/3grouping function after adensity/3call. Instead use the:color_byoption, which will ensure that the proper settings are applied to the underlying transformation.Calling
color_by/3would produce this graph:Tucan.density(:penguins, "Body Mass (g)") |> Tucan.color_by("Species")In the above case the density function has been calculated on the complete dataset and you cannot color by the
Species. Instead you should use the:color_byoption which would calculate the density function per group:Tucan.density(:penguins, "Body Mass (g)", color_by: "Species", fill_opacity: 0.2)Alternatively you should use the
:groupbyoption in order to group the density transform by theSpeciesfield and then apply thecolor_by/3function:Tucan.density(:penguins, "Body Mass (g)", groupby: ["Species"], fill_opacity: 0.2) |> Tucan.color_by("Species")
See also histogram/3.
Options
:bandwidth(float/0) - The bandwidth (standard deviation) of the Gaussian kernel. If unspecified or set to zero, the bandwidth value is automatically estimated from the input data using Scott’s rule.:counts(boolean/0) - A boolean flag indicating if the output values should be probability estimates (false) or smoothed counts (true). The default value isfalse.:cumulative(boolean/0) - A boolean flag indicating whether to produce density estimates (false) or cumulative density estimates (true). The default value isfalse.:extent- A[min, max]domain from which to sample the distribution. If unspecified, the extent will be determined by the observed minimum and maximum values of the density value field.:groupby(list ofString.t/0) - The data fields to group by. If not specified, a single group containing all data objects will be used. This is applied only on the density transform.In most cases you only need to set
color_bywhich will automatically handle the density transform grouping. Usegroupbyonly if you want to manually post-process the generated specification, or if you want to apply grouping by more than one variable.If both
groupbyandcolor_byare set then onlygroupbyis used for grouping the density transform andcolor_byis used for encoding the color.:maxsteps(integer/0) - The maximum number of samples to take along the extent domain for plotting the density. The default value is200.:minsteps(integer/0) - The minimum number of samples to take along the extent domain for plotting the density. The default value is25.:steps(integer/0) - The exact number of samples to take along the extent domain for plotting the density. If specified, overrides both minsteps and maxsteps to set an exact number of uniform samples. Potentially useful in conjunction with a fixed extent to ensure consistent sample points for stacked densities.
Data Grouping Options
:color_by(String.t/0) - If set a data field that will be used for coloring the data. It is considered:nominalby default.
Styling Options
:clip(boolean/0) - Whether a mark will be clipped to the enclosing group’s width and height.:fill_opacity(number/0) - The fill opacity of the plotted elements. The default value is1.:height(integer/0) - Height of the image:orient(atom/0) - The plot's orientation, can be either:horizontalor:vertical. The default value is:horizontal.:title(String.t/0) - The title of the graph:width(integer/0) - Width of the image
Encodings Custom Options
All Tucan plots are building a VegaLite specification based on some sane
default parameters. Through these encodings options you are free to set any
vega-lite supported option to any encoding channel of the plot.
Notice that if set they will be merged with any option set by the plot. Since they have a higher precedence they may override the default settings and affect the generated plot.
You can set an arbitrary keyword list. Notice that the contents are not validated.
:color(keyword/0) - Extra vega lite options for the:colorencoding. The default value is[].:x(keyword/0) - Extra vega lite options for the:xencoding. The default value is[].:y(keyword/0) - Extra vega lite options for the:yencoding. The default value is[].
Interactivity Options
:tooltip(boolean() | :data | :encoding) - The tooltip text string to show upon mouse hover or an object defining which fields should the tooltip be derived from. Can be one of the following::encoding- all fields from encoding are used:data- all fields of the highlighted data point are usedtrue- same as:encodingfalse,nil- no tooltip is used
Examples
Tucan.density(:penguins, "Body Mass (g)")It is a common use case to compare the density of several groups in a dataset. Several options exist to do so. You can plot all items on the same chart, using transparency and annotation to make the comparison possible.
Tucan.density(:penguins, "Body Mass (g)", color_by: "Species", fill_opacity: 0.5)You can also combine it with facet_by/4 in order to draw a different plot for each value
of the grouping variable. Notice that we need to set the :groupby variable in order
to correctly calculate the density plot per field's value.
Tucan.density(:penguins, "Body Mass (g)", groupby: ["Species"])
|> Tucan.color_by("Species")
|> Tucan.facet_by(:column, "Species")You can control the smoothing by setting a specific bandwidth value (if not set it is
automatically calculated by vega lite):
Tucan.density(:penguins, "Body Mass (g)", color_by: "Species", bandwidth: 20.0, fill_opacity: 0.5)You can plot a cumulative density distribution by setting the :cumulative option to true:
Tucan.density(:penguins, "Body Mass (g)", cumulative: true)or calculate a separate cumulative distribution for each group:
Tucan.density(:penguins, "Body Mass (g)", cumulative: true, color_by: "Species")
|> Tucan.facet_by(:column, "Species")@spec density_heatmap( plotdata :: plotdata(), x :: binary(), y :: binary(), opts :: keyword() ) :: VegaLite.t()
Draws a density heatmap.
A density heatmap is a bivariate histogram, e.g. the x, y data are binned
within rectangles that tile the plot and then the count of observations within
each rectangle is shown with the fill color.
By default the count of observations within each rectangle is encoded, but you
can calculate the statistic of any field and use it instead.
Density heatmaps are a powerful visualization tool that find their best use cases in situations where you need to explore and understand the distribution and concentration of data points in a two-dimensional space. They are particularly effective when dealing with large datasets, allowing you to uncover patterns, clusters, and trends that might be difficult to discern in raw data.
Options
:aggregate(atom/0) - The statistic that will be used for aggregating the observations within a bin. Thezfield must be set ifaggregateis set.:z(String.t/0) - If set corresponds to the field that will be used for calculating the color fo the bin using the provided aggregate. If not set (the default behaviour) the count of observations are used for coloring the bin.
Styling Options
:clip(boolean/0) - Whether a mark will be clipped to the enclosing group’s width and height.:fill_opacity(number/0) - The fill opacity of the plotted elements. The default value is1.:height(integer/0) - Height of the image:title(String.t/0) - The title of the graph:width(integer/0) - Width of the image
Encodings Custom Options
All Tucan plots are building a VegaLite specification based on some sane
default parameters. Through these encodings options you are free to set any
vega-lite supported option to any encoding channel of the plot.
Notice that if set they will be merged with any option set by the plot. Since they have a higher precedence they may override the default settings and affect the generated plot.
You can set an arbitrary keyword list. Notice that the contents are not validated.
:color(keyword/0) - Extra vega lite options for the:colorencoding. The default value is[].:x(keyword/0) - Extra vega lite options for the:xencoding. The default value is[].:y(keyword/0) - Extra vega lite options for the:yencoding. The default value is[].
Interactivity Options
:tooltip(boolean() | :data | :encoding) - The tooltip text string to show upon mouse hover or an object defining which fields should the tooltip be derived from. Can be one of the following::encoding- all fields from encoding are used:data- all fields of the highlighted data point are usedtrue- same as:encodingfalse,nil- no tooltip is used
Examples
Let's start with a default density heatmap on the penguins dataset:
Tucan.density_heatmap(:penguins, "Beak Length (mm)", "Beak Depth (mm)")You can summarize over another field:
Tucan.density_heatmap(:penguins, "Beak Length (mm)", "Beak Depth (mm)",
z: "Body Mass (g)",
aggregate: :mean
)@spec donut( plotdata :: plotdata(), field :: binary(), category :: binary(), opts :: keyword() ) :: VegaLite.t()
Draw a donut chart.
A donut chart is a circular visualization that resembles a pie chart but features a hole at its center. This central hole creates a donut shape, distinguishing it from traditional pie charts.
This is a wrapper around pie/4 that sets by default the :inner_radius.
Options
See pie/4
Examples
Tucan.donut(:barley, "yield", "site", aggregate: :sum, tooltip: true)
|> Tucan.facet_by(:column, "year", type: :nominal)@spec heatmap( plotdata :: plotdata(), x :: binary(), y :: binary(), color :: nil | binary(), opts :: keyword() ) :: VegaLite.t()
Returns the specification of a heatmap.
A heatmap is a graphical representation of data where the individual values contained in a matrix are represented as colors.
It expects two categorical fields x, y which will be used for the axes
and a numerical field color. If color is nil then the color represents
the count of the observations for each x, y.
If an :aggregate is set this statistic will be used for encoding the color.
If no :aggregate is set the color encodes by default the :mean of the
data.
Options
:aggregate(atom/0) - The statistic that will be used for aggregating the observations within a heatmap tile. Defaults to:meanwhich in case of single data will encode the value of thecolordata field.Ignored if
:coloris set tonil.:annotate(boolean/0) - If set totruethen the values of each cell will be included in the plot. The default value isfalse.
Styling Options
:clip(boolean/0) - Whether a mark will be clipped to the enclosing group’s width and height.:color_scheme(atom/0) - The colorscheme to use, for supported colorschemes checkTucan.Scale. Notice that this is just a helper option for easily setting color schemes. If you need to set specific colors or customize the scheme, useTucan.Scale.set_color_scheme/3.:fill_opacity(number/0) - The fill opacity of the plotted elements. The default value is1.:height(integer/0) - Height of the image:title(String.t/0) - The title of the graph:width(integer/0) - Width of the image
Encodings Custom Options
All Tucan plots are building a VegaLite specification based on some sane
default parameters. Through these encodings options you are free to set any
vega-lite supported option to any encoding channel of the plot.
Notice that if set they will be merged with any option set by the plot. Since they have a higher precedence they may override the default settings and affect the generated plot.
You can set an arbitrary keyword list. Notice that the contents are not validated.
:color(keyword/0) - Extra vega lite options for the:colorencoding. The default value is[].:text(keyword/0) - Extra vega lite options for the:textencoding. The default value is[].:x(keyword/0) - Extra vega lite options for the:xencoding. The default value is[].:y(keyword/0) - Extra vega lite options for the:yencoding. The default value is[].
Interactivity Options
:tooltip(boolean() | :data | :encoding) - The tooltip text string to show upon mouse hover or an object defining which fields should the tooltip be derived from. Can be one of the following::encoding- all fields from encoding are used:data- all fields of the highlighted data point are usedtrue- same as:encodingfalse,nil- no tooltip is used
Examples
A simple heatmap of two categorical variables, using a third one for the color values.
data = [
%{"x" => "A", "y" => "K", "value" => 0.5},
%{"x" => "A", "y" => "L", "value" => 1.5},
%{"x" => "A", "y" => "M", "value" => 4.5},
%{"x" => "B", "y" => "K", "value" => 1.5},
%{"x" => "B", "y" => "L", "value" => 2.5},
%{"x" => "B", "y" => "M", "value" => 0.5},
%{"x" => "C", "y" => "K", "value" => -1.5},
%{"x" => "C", "y" => "L", "value" => 5.5},
%{"x" => "C", "y" => "M", "value" => 1.5}
]
Tucan.heatmap(data, "x", "y", "value", width: 200, height: 200)You can change the color scheme:
Tucan.heatmap(:glue, "Task", "Model", "Score", color_scheme: :redyellowgreen, tooltip: true)Heatmaps are also useful for visualizing temporal data. Let's use a heatmap to examine how Seattle's max temperature changes over the year. On the x-axis we will encode the days of the month along the x-axis, and the months on the y-axis. We will aggregate over the max temperature for the color field. (example borrowed from here)
Tucan.heatmap(:weather, "date", "date", "temp_max",
x: [type: :ordinal, time_unit: :date],
y: [type: :ordinal, time_unit: :month],
tooltip: true
)
|> Tucan.Scale.set_color_scheme(:redyellowblue, reverse: true)
|> Tucan.Axes.set_x_title("Day")
|> Tucan.Axes.set_y_title("Month")
|> Tucan.Legend.set_title(:color, "Avg Max Temp")
|> Tucan.set_title("Heatmap of Avg Max Temperatures in Seattle (2012-2015)")You can enable annotations by setting the :annotate flag:
Tucan.heatmap(:weather, "date", "date", "temp_max",
annotate: true,
x: [type: :ordinal, time_unit: :date],
y: [type: :ordinal, time_unit: :month],
text: [format: ".1f"],
tooltip: true,
width: 800
)
|> Tucan.Scale.set_color_scheme(:redyellowblue, reverse: true)
|> Tucan.Axes.set_x_title("Day")
|> Tucan.Axes.set_y_title("Month")
|> Tucan.Legend.set_title(:color, "Avg Max Temp")
|> Tucan.set_title("Heatmap of Avg Max Temperatures in Seattle (2012-2015)")@spec histogram(plotdata :: plotdata(), field :: binary(), opts :: keyword()) :: VegaLite.t()
Plots a histogram.
See also density/3
Options
:extent- A two-element ([min, max]) array indicating the range of desired bin values.:maxbins(integer/0) - Maximum number of bins.:orient- Histogram's orientation. It specifies the axis along which the field values are plotted. The default value is:horizontal.:relative(boolean/0) - If set a relative frequency histogram is generated. The default value isfalse.:stacked(boolean/0) - If set it will stack the group histograms instead of layering one over another. Valid only if a semantic grouping has been applied.:step- An exact step size to use between bins. If provided, options such asmaxbinswill be ignored.
Data Grouping Options
:color_by(String.t/0) - The field to group observations by. This will used for coloring the histogram if set.
Styling Options
:clip(boolean/0) - Whether a mark will be clipped to the enclosing group’s width and height.:fill_opacity(number/0) - The fill opacity of the plotted elements. The default value is1.:height(integer/0) - Height of the image:title(String.t/0) - The title of the graph:width(integer/0) - Width of the image
Encodings Custom Options
All Tucan plots are building a VegaLite specification based on some sane
default parameters. Through these encodings options you are free to set any
vega-lite supported option to any encoding channel of the plot.
Notice that if set they will be merged with any option set by the plot. Since they have a higher precedence they may override the default settings and affect the generated plot.
You can set an arbitrary keyword list. Notice that the contents are not validated.
:color(keyword/0) - Extra vega lite options for the:colorencoding. The default value is[].:x(keyword/0) - Extra vega lite options for the:xencoding. The default value is[].:x2(keyword/0) - Extra vega lite options for the:x2encoding. The default value is[].:y(keyword/0) - Extra vega lite options for the:yencoding. The default value is[].
Interactivity Options
:tooltip(boolean() | :data | :encoding) - The tooltip text string to show upon mouse hover or an object defining which fields should the tooltip be derived from. Can be one of the following::encoding- all fields from encoding are used:data- all fields of the highlighted data point are usedtrue- same as:encodingfalse,nil- no tooltip is used
Examples
Histogram of Horsepower
Tucan.histogram(:cars, "Horsepower")You can flip the plot by setting the :orient option to :vertical:
Tucan.histogram(:cars, "Horsepower", orient: :vertical)By setting the :relative flag you can get a relative frequency histogram:
Tucan.histogram(:cars, "Horsepower", relative: true)You can increase the number of bins by settings the maxbins or the step options:
Tucan.histogram(:cars, "Horsepower", step: 5)You can draw multiple histograms by grouping the observations by a second categorical variable:
Tucan.histogram(:cars, "Horsepower", color_by: "Origin", fill_opacity: 0.5)By default the histograms are plotted layered, but you can also stack them:
Tucan.histogram(:cars, "Horsepower", color_by: "Origin", fill_opacity: 0.5, stacked: true)or you can facet it, in order to make the histograms more clear:
histograms =
Tucan.histogram(:cars, "Horsepower", color_by: "Origin", tooltip: true)
|> Tucan.facet_by(:column, "Origin")
relative_histograms =
Tucan.histogram(:cars, "Horsepower", relative: true, color_by: "Origin", tooltip: true)
|> Tucan.facet_by(:column, "Origin")
Tucan.vconcat([histograms, relative_histograms])@spec lineplot(plotdata :: plotdata(), x :: field(), y :: field(), opts :: keyword()) :: VegaLite.t()
Draw a line plot between x and y
Both x and y are considered numerical variables.
Options
:interpolate(binary/0) - The line interpolation method to use for line and area marks. One of the following:"linear"- piecewise linear segments, as in a poly-line."linear-closed"- close the linear segments to form a polygon."step"- alternate between horizontal and vertical segments, as in a step function."step-before"- alternate between vertical and horizontal segments, as in a step function."step-after"- alternate between horizontal and vertical segments, as in a step function."basis"- a B-spline, with control point duplication on the ends."basis-open"- an open B-spline; may not intersect the start or end."basis-closed"- a closed B-spline, as in a loop."cardinal"- a Cardinal spline, with control point duplication on the ends."cardinal-open"- an open Cardinal spline; may not intersect the start or end, but will intersect other control points."cardinal-closed"- a closed Cardinal spline, as in a loop."bundle"- equivalent to basis, except the tension parameter is used to straighten the spline."monotone"- cubic interpolation that preserves monotonicity in y.
:tension(number/0) - Depending on the interpolation type, sets the tension parameter
Data Grouping Options
:color_by(String.t/0) - If set a data field that will be used for coloring the data. It is considered:nominalby default.:group_by(String.t/0) - A field to group by the lines without affecting the style of it.
Styling Options
:clip(boolean/0) - Whether a mark will be clipped to the enclosing group’s width and height.:fill_opacity(number/0) - The fill opacity of the plotted elements. The default value is1.:filled(boolean/0) - Whether the points will be filled or not. Valid only if:pointsis set. The default value istrue.:height(integer/0) - Height of the image:line_color(String.t/0) - The color of the line:points(boolean/0) - Whether points will be included in the chart. The default value isfalse.:title(String.t/0) - The title of the graph:width(integer/0) - Width of the image
Encodings Custom Options
All Tucan plots are building a VegaLite specification based on some sane
default parameters. Through these encodings options you are free to set any
vega-lite supported option to any encoding channel of the plot.
Notice that if set they will be merged with any option set by the plot. Since they have a higher precedence they may override the default settings and affect the generated plot.
You can set an arbitrary keyword list. Notice that the contents are not validated.
:color(keyword/0) - Extra vega lite options for the:colorencoding. The default value is[].:x(keyword/0) - Extra vega lite options for the:xencoding. The default value is[].:y(keyword/0) - Extra vega lite options for the:yencoding. The default value is[].
Interactivity Options
:tooltip(boolean() | :data | :encoding) - The tooltip text string to show upon mouse hover or an object defining which fields should the tooltip be derived from. Can be one of the following::encoding- all fields from encoding are used:data- all fields of the highlighted data point are usedtrue- same as:encodingfalse,nil- no tooltip is used
Examples
Plotting a simple line chart of Google stock price over time. Notice how we change the
x axis type from the default (:quantitative) to :temporal using the generic
:x channel configuration option:
Tucan.lineplot(:stocks, "date", "price", x: [type: :temporal])
|> VegaLite.transform(filter: "datum.symbol==='GOOG'")You could plot all stocks of the dataset with different colors by setting the :color_by
option. If you do not want to color lines differently, you can pass the :group_by option
instead of :color_by:
left = Tucan.lineplot(:stocks, "date", "price", x: [type: :temporal], color_by: "symbol")
right = Tucan.lineplot(:stocks, "date", "price", x: [type: :temporal], group_by: "symbol")
Tucan.hconcat([left, right])You can also overlay the points by setting the :points and :filled opts. Notice
that below we plot by year and aggregating the y values:
filled_points =
Tucan.lineplot(:stocks, "date", "price",
x: [type: :temporal, time_unit: :year],
y: [aggregate: :mean],
color_by: "symbol",
points: true,
tooltip: true,
width: 300
)
stroked_points =
Tucan.lineplot(:stocks, "date", "price",
x: [type: :temporal, time_unit: :year],
y: [aggregate: :mean],
color_by: "symbol",
points: true,
filled: false,
tooltip: true,
width: 300
)
Tucan.hconcat([filled_points, stroked_points])You can use various interpolation methods. Some examples follow:
plots =
for interpolation <- ["linear", "step", "cardinal", "monotone"] do
Tucan.lineplot(:stocks, "date", "price",
x: [type: :temporal, time_unit: :year],
y: [aggregate: :mean],
color_by: "symbol",
interpolate: interpolation
)
|> Tucan.set_title(interpolation)
end
VegaLite.new(columns: 2)
|> Tucan.concat(plots)@spec pie( plotdata :: plotdata(), field :: binary(), category :: binary(), opts :: keyword() ) :: VegaLite.t()
Draws a pie chart.
A pie chart is a circle divided into sectors that each represents a proportion
of the whole. The field specifies the data column that contains the proportions
of each category. The chart will be colored by the category field.
Avoid using pie charts
Despite it's popularity pie charts should rarely be used. Pie charts are best suited for displaying a small number of categories and can make it challenging to accurately compare data. They rely on angle perception, which can lead to misinterpretation, and lack the precision offered by other charts like bar charts or line charts.
Instead, opt for alternatives such as bar charts for straightforward comparisons, stacked area charts for cumulative effects.
The following example showcases the limitations of a pie chart, compared to a bar chart:
alias VegaLite, as: Vl data = [ %{value: 30, category: "A"}, %{value: 33, category: "B"}, %{value: 38, category: "C"} ] pie = Tucan.pie(data, "value", "category") bar = Tucan.bar(data, "category", "value", orient: :vertical) Tucan.hconcat([pie, bar]) |> Tucan.set_title("Pie vs Bar chart", anchor: :middle, offset: 15)
Options
:aggregate(atom/0) - The statistic to use (if any) for aggregating values per pie slice (e.g.:mean).:inner_radius(integer/0) - The inner radius in pixels.0for a pie chart,> 0for a donut chart. If not set it defaults to 0
Styling Options
:clip(boolean/0) - Whether a mark will be clipped to the enclosing group’s width and height.:fill_opacity(number/0) - The fill opacity of the plotted elements. The default value is1.:height(integer/0) - Height of the image:title(String.t/0) - The title of the graph:width(integer/0) - Width of the image
Encodings Custom Options
All Tucan plots are building a VegaLite specification based on some sane
default parameters. Through these encodings options you are free to set any
vega-lite supported option to any encoding channel of the plot.
Notice that if set they will be merged with any option set by the plot. Since they have a higher precedence they may override the default settings and affect the generated plot.
You can set an arbitrary keyword list. Notice that the contents are not validated.
:color(keyword/0) - Extra vega lite options for the:colorencoding. The default value is[].:theta(keyword/0) - Extra vega lite options for the:thetaencoding. The default value is[].
Interactivity Options
:tooltip(boolean() | :data | :encoding) - The tooltip text string to show upon mouse hover or an object defining which fields should the tooltip be derived from. Can be one of the following::encoding- all fields from encoding are used:data- all fields of the highlighted data point are usedtrue- same as:encodingfalse,nil- no tooltip is used
Examples
Tucan.pie(:barley, "yield", "site", aggregate: :sum, tooltip: true)
|> Tucan.facet_by(:column, "year", type: :nominal)@spec punchcard( plotdata :: plotdata(), x :: binary(), y :: binary(), size :: nil | binary(), opts :: keyword() ) :: VegaLite.t()
Returns the specification of a punch card plot.
A punch card plot is similar to a heatmap but instead of color the third dimension is encoded by the size of bubbles.
See also heatmap/5.
Options
:aggregate(atom/0) - The statistic that will be used for aggregating the observations within a heatmap tile. Defaults to:meanwhich in case of single data will encode the value of thecolordata field.Ignored if
:coloris set tonil.:annotate(boolean/0) - If set totruethen the values of each cell will be included in the plot. The default value isfalse.
Styling Options
:clip(boolean/0) - Whether a mark will be clipped to the enclosing group’s width and height.:color_scheme(atom/0) - The colorscheme to use, for supported colorschemes checkTucan.Scale. Notice that this is just a helper option for easily setting color schemes. If you need to set specific colors or customize the scheme, useTucan.Scale.set_color_scheme/3.:fill_opacity(number/0) - The fill opacity of the plotted elements. The default value is1.:height(integer/0) - Height of the image:title(String.t/0) - The title of the graph:width(integer/0) - Width of the image
Encodings Custom Options
All Tucan plots are building a VegaLite specification based on some sane
default parameters. Through these encodings options you are free to set any
vega-lite supported option to any encoding channel of the plot.
Notice that if set they will be merged with any option set by the plot. Since they have a higher precedence they may override the default settings and affect the generated plot.
You can set an arbitrary keyword list. Notice that the contents are not validated.
:size(keyword/0) - Extra vega lite options for the:sizeencoding. The default value is[].:x(keyword/0) - Extra vega lite options for the:xencoding. The default value is[].:y(keyword/0) - Extra vega lite options for the:yencoding. The default value is[].
Interactivity Options
:tooltip(boolean() | :data | :encoding) - The tooltip text string to show upon mouse hover or an object defining which fields should the tooltip be derived from. Can be one of the following::encoding- all fields from encoding are used:data- all fields of the highlighted data point are usedtrue- same as:encodingfalse,nil- no tooltip is used
Examples
Tucan.punchcard(:weather, "date", "date", "temp_max",
tooltip: true,
x: [type: :ordinal, time_unit: :date],
y: [type: :ordinal, time_unit: :month]
)
|> Tucan.Axes.set_x_title("Day")
|> Tucan.Axes.set_y_title("Month")
|> Tucan.set_title("Punch card of Avg Max Temperatures in Seattle (2012-2015)")You can add a fourth dimension by coloring the plot by a fourth variable. Notice how
we use Tucan.Scale.set_color_scheme/3 to apply a semantically reasonable coloring and
Tucan.Legend.set_orientation/3 to change the default position of the two legends.
Tucan.punchcard(:weather, "date", "date", "precipitation",
tooltip: true,
x: [type: :ordinal, time_unit: :date],
y: [type: :ordinal, time_unit: :month]
)
# we need to set recursive to true since this is a layered plot
|> Tucan.color_by("temp_max", aggregate: :mean, recursive: true)
|> Tucan.Scale.set_color_scheme(:redyellowblue, reverse: true)
|> Tucan.Axes.set_x_title("Day")
|> Tucan.Axes.set_y_title("Month")
|> Tucan.Legend.set_orientation(:color, "bottom")
|> Tucan.Legend.set_orientation(:size, "bottom")@spec scatter(plotdata :: plotdata(), x :: binary(), y :: binary(), opts :: keyword()) :: VegaLite.t()
Returns the specification of a scatter plot with possibility of several semantic groupings.
Both x and y must be :quantitative.
Semantic groupings
The relationship between
xandycan be shown for different subsets of the data using thecolor_by,size_byandshape_byparameters. This is equivalent to calling the corresponding functions after ascatter/4call.These parameters control what visual semantics are used to identify the different subsets. It is possible to show up to three dimensions independently by using all three semantic types, but this style of plot can be hard to interpret and is often ineffective.
Tucan.scatter(:tips, "total_bill", "tip", color_by: "day", shape_by: "sex", size_by: "size" )The above is equivalent to calling:
Tucan.scatter(:tips, "total_bill", "tip") |> Tucan.color_by("day", type: :nominal) |> Tucan.shape_by("sex", type: :nominal) |> Tucan.size_by("size", type: :quantitative)Using redundant semantics (i.e. both color and shape for the same variable) can be helpful for making graphics more accessible.
Tucan.scatter(:tips, "total_bill", "tip", color_by: "day", shape_by: "day" )
Options
Data Grouping Options
:color_by(String.t/0) - If set a data field that will be used for coloring the data. It is considered:nominalby default.:shape_by(String.t/0) - If set a data field that will be used for setting the shape of the data points. It is considered:nominalby default.:size_by(String.t/0) - If set a data field that will be used for controlling the size of the data points. It is considered:quantitativeby default.
Styling Options
:clip(boolean/0) - Whether a mark will be clipped to the enclosing group’s width and height.:fill_opacity(number/0) - The fill opacity of the plotted elements. The default value is1.:filled(boolean/0) - Whether the mark will be filled or not:height(integer/0) - Height of the image:point_color(String.t/0) - The color of the points:point_shape- Shape of the point marks. Circle by default.:point_size(pos_integer/0) - The pixel area of the marks. Note that this value sets the area of the symbol; the side lengths will increase with the square root of this value.:title(String.t/0) - The title of the graph:width(integer/0) - Width of the image
Encodings Custom Options
All Tucan plots are building a VegaLite specification based on some sane
default parameters. Through these encodings options you are free to set any
vega-lite supported option to any encoding channel of the plot.
Notice that if set they will be merged with any option set by the plot. Since they have a higher precedence they may override the default settings and affect the generated plot.
You can set an arbitrary keyword list. Notice that the contents are not validated.
:color(keyword/0) - Extra vega lite options for the:colorencoding. The default value is[].:shape(keyword/0) - Extra vega lite options for the:shapeencoding. The default value is[].:size(keyword/0) - Extra vega lite options for the:sizeencoding. The default value is[].:x(keyword/0) - Extra vega lite options for the:xencoding. The default value is[].:y(keyword/0) - Extra vega lite options for the:yencoding. The default value is[].
Interactivity Options
:tooltip(boolean() | :data | :encoding) - The tooltip text string to show upon mouse hover or an object defining which fields should the tooltip be derived from. Can be one of the following::encoding- all fields from encoding are used:data- all fields of the highlighted data point are usedtrue- same as:encodingfalse,nil- no tooltip is used
Examples
We will use the
:tipsdataset throughout the following examples.
Drawing a scatter plot between two variables:
Tucan.scatter(:tips, "total_bill", "tip")You can modify the look of the plot by setting various styling options:
Tucan.scatter(:tips, "total_bill", "tip",
point_color: "red",
point_shape: "triangle-up",
point_size: 10
)You can combine it with color_by/3 to color code the points with respect to
another variable:
Tucan.scatter(:tips, "total_bill", "tip")
|> Tucan.color_by("time")Assigning the same variable to shape_by/3 will also vary the markers and create a
more accessible plot:
Tucan.scatter(:tips, "total_bill", "tip", width: 400)
|> Tucan.color_by("time")
|> Tucan.shape_by("time")Assigning color_by/3 and shape_by/3 to different variables will vary colors and
markers independently:
Tucan.scatter(:tips, "total_bill", "tip", width: 400)
|> Tucan.color_by("day")
|> Tucan.shape_by("time")You can also color the points by a numeric variable, the semantic mapping will be quantitative and will use a different default palette:
Tucan.scatter(:tips, "total_bill", "tip", width: 400)
|> Tucan.color_by("size", type: :quantitative)A numeric variable can also be assigned to size to apply a semantic mapping to the areas of the points:
Tucan.scatter(:tips, "total_bill", "tip", width: 400, tooltip: :data)
|> Tucan.color_by("size", type: :quantitative)
|> Tucan.size_by("size", type: :quantitative)You can also combine it with facet_by/3 in order to group within additional
categorical variables, and plot them across multiple subplots.
Tucan.scatter(:tips, "total_bill", "tip", width: 300)
|> Tucan.color_by("day")
|> Tucan.shape_by("day")
|> Tucan.facet_by(:column, "time")You can also apply faceting on more than one variables, both horizontally and vertically:
Tucan.scatter(:tips, "total_bill", "tip", width: 300)
|> Tucan.color_by("day")
|> Tucan.shape_by("day")
|> Tucan.size_by("size")
|> Tucan.facet_by(:column, "time")
|> Tucan.facet_by(:row, "sex")@spec step(plotdata :: plotdata(), x :: field(), y :: field(), opts :: keyword()) :: VegaLite.t()
Returns the specification of a step chart.
This is a simple wrapper around lineplot/4 with :interpolate set by default
to "step". If :interpolate is set to any of step, step-before, step-after it
will be used. In any other case defaults to step.
Options
Check lineplot/4
Examples
Tucan.step(:stocks, "date", "price", color_by: "symbol", width: 300, x: [type: :temporal])
|> Tucan.Scale.set_y_scale(:log)@spec streamgraph( plotdata :: plotdata(), x :: field(), y :: field(), group :: field(), opts :: keyword() ) :: VegaLite.t()
Returns the specification of a streamgraph.
This is a simple wrapper around area/4 with :mode set by default
to :streamgraph. Any value set to the :mode option will be ignored.
A grouping field must also be provided which will be set as :color_by to
the area chart.
Options
Check area/4
Examples
Tucan.streamgraph(:stocks, "date", "price", "symbol",
width: 300,
x: [type: :temporal],
tooltip: true
)@spec stripplot(plotdata :: plotdata(), field :: binary(), opts :: keyword()) :: VegaLite.t()
Draws a strip plot (categorical scatterplot).
A strip plot is a single-axis scatter plot used to visualize the distribution of a numerical field. The values are plotted as dots or ticks along one axis, so the dots with the same value may overlap.
You can use the :jitter mode for a better view of overlapping points. In this
case points are randomly shifted along with other axis, which has no meaning in
itself data-wise.
Typically several strip plots are placed side by side to compare the distribution of a numerical value among several categories.
Options
:group(String.t/0) - A field to be used for grouping the strip plot. If not set the plot will be one dimensional.:style- The style of the plot. Can be one of the following::tick- use ticks for each data point:point- use points for each data point:jitter- use points but also apply some jittering across the other axis
Use
:jitterin case of many data points in order to avoid overlaps.The default value is
:tick.
Data Grouping Options
:color_by(String.t/0) - If set a data field that will be used for coloring the data. It is considered:nominalby default.
Styling Options
:clip(boolean/0) - Whether a mark will be clipped to the enclosing group’s width and height.:fill_opacity(number/0) - The fill opacity of the plotted elements. The default value is1.:height(integer/0) - Height of the image:orient(atom/0) - The plot's orientation, can be either:horizontalor:vertical. The default value is:horizontal.:title(String.t/0) - The title of the graph:width(integer/0) - Width of the image
Encodings Custom Options
All Tucan plots are building a VegaLite specification based on some sane
default parameters. Through these encodings options you are free to set any
vega-lite supported option to any encoding channel of the plot.
Notice that if set they will be merged with any option set by the plot. Since they have a higher precedence they may override the default settings and affect the generated plot.
You can set an arbitrary keyword list. Notice that the contents are not validated.
:color(keyword/0) - Extra vega lite options for the:colorencoding. The default value is[].:x(keyword/0) - Extra vega lite options for the:xencoding. The default value is[].:y(keyword/0) - Extra vega lite options for the:yencoding. The default value is[].:y_offset(keyword/0) - Extra vega lite options for the:y_offsetencoding. The default value is[].
Interactivity Options
:tooltip(boolean() | :data | :encoding) - The tooltip text string to show upon mouse hover or an object defining which fields should the tooltip be derived from. Can be one of the following::encoding- all fields from encoding are used:data- all fields of the highlighted data point are usedtrue- same as:encodingfalse,nil- no tooltip is used
Internal
VegaLiterepresentationIf style is set to
:tickthe followingVegaLiterepresentation is generated:Vl.new() |> Vl.mark(:tick) |> Vl.encode_field(:x, field, type: :quantitative) |> Vl.encode_field(:y, opts[:group], type: :nominal)If style is set to
:jitterthen a transform is added to generate Gaussian jitter using the Box-Muller transform, and they_offsetis also encoded based on this:Vl.new() |> Vl.mark(:point) |> Vl.transform(calculate: "sqrt(-2*log(random()))*cos(2*PI*random())", as: "jitter") |> Vl.encode_field(:x, field, type: :quantitative) |> Vl.encode_field(:y, opts[:group], type: :nominal) |> Vl.encode_field(:y_offset, "jitter", type: :quantitative)
Examples
Assigning a single numeric variable shows the univariate distribution. The default
style is the :tick:
Tucan.stripplot(:tips, "total_bill")For very dense distribution it makes more sense to use the :jitter style in order
to reduce overlapping points:
Tucan.stripplot(:tips, "total_bill", style: :jitter, height: 30, width: 300)You can set the :group option in order to add a second dimension. Notice that
the field must be categorical.
Tucan.stripplot(:tips, "total_bill", group: "day", style: :jitter)The plot would be more clear if you also colored the points with the same field:
Tucan.stripplot(:tips, "total_bill", group: "day", style: :jitter)
|> Tucan.color_by("day")Or you can color by a distinct variable to show a multi-dimensional relationship:
Tucan.stripplot(:tips, "total_bill", group: "day", style: :jitter)
|> Tucan.color_by("sex")or you can color by a numerical variable:
Tucan.stripplot(:tips, "total_bill", group: "day", style: :jitter)
|> Tucan.color_by("size", type: :ordinal)You could draw the same with points but without jittering:
Tucan.stripplot(:tips, "total_bill", group: "day", style: :point)
|> Tucan.color_by("sex")or with ticks which is the default one:
Tucan.stripplot(:tips, "total_bill", group: "day", style: :tick)
|> Tucan.color_by("sex")You can set the :orient flag to :vertical to change the orientation:
Tucan.stripplot(:tips, "total_bill", group: "day", style: :jitter, orient: :vertical)
|> Tucan.color_by("sex")Composite Plots
@spec jointplot(plotdata :: plotdata(), x :: field(), y :: field(), opts :: keyword()) :: VegaLite.t()
Returns the specification of a jointplot.
A jointplot is a plot of two numerical variables along with marginal univariate graphs. If no options are set the joint is a scatter plot and the marginal are the histograms of the two variables.
Marginal plots dimensions
By default a jointplot will have a square shape, e.g. it will have the same width and height. The
:widthoption affects the width of the central (joint) plot.For the marginal distributions you can the
:ratiooption which specifies the ratio of joint axes height to marginal axes height.
Options
:joint- The plot type to be used for the main (joint) plot. Can be one of:scatterand:density_heatmap. The default value is:scatter.:joint_opts(keyword/0) - Arbitrary options list for the joint plot. The supported options depend on the selected:jointtype. The default value is[].:marginal- The plot type to be used for the marginal plots. Can be one of:histogramand:density. The default value is:histogram.:marginal_opts(keyword/0) - Arbitrary options list for the marginal plots. The supported options depend on the selected:marginaltype. The default value is[].:ratio(float/0) - The ratio of the marginal plots secondary dimension with respect to the joint plot dimension. The default value is0.45.
Data Grouping Options
:color_by(String.t/0) - If set a data field that will be used for coloring the data. It is considered:nominalby default.
Styling Options
:fill_opacity(number/0) - The fill opacity of the plotted elements. The default value is1.:spacing(pos_integer/0) - The spacing between the marginals and the joint plot. The default value is15.:title(String.t/0) - The title of the graph:width(integer/0) - The dimension of the central (joint) plot. The same value is used for both the width and height of the plot. The default value is200.
Examples
A simple joint plot between two variables.
Tucan.jointplot(:iris, "petal_width", "petal_length", width: 200)You can also pass :color_by to apply a semantic grouping. If set it will be
applied both to the joint and the marginal plots.
Tucan.jointplot(
:iris,
"petal_width",
"petal_length",
color_by: "species",
fill_opacity: 0.5,
width: 200
)You can change the type of the join plot and the marginal distributions:
Tucan.jointplot(
:penguins,
"Beak Length (mm)",
"Beak Depth (mm)",
joint: :density_heatmap,
marginal: :density,
ratio: 0.3
)@spec pairplot(plotdata :: plotdata(), fields :: [binary()], opts :: keyword()) :: VegaLite.t()
Plot pairwise relationships in a dataset.
This function expects an array of fields to be provided. A grid will be created
where each numeric variable in fields will be shared across the y-axes across
a single row and the x-axes across a single column.
Numerical field types
Notice that currently
pairplot/3works only with numerical (:quantitative) variables. If you need to create a pair plot containing other variable types you need to manually build the grid using theVegaLiteconcatenation operations.
Options
:diagonal- The plot type to be used for the diagonal subplots. Can be one on:scatter,:densityand:histogram. The default value is:scatter.:plot_fn(function of arity 3) - An optional function for customizing the look any subplot. It expects a function with the following signature:(vl :: VegaLite.t(), row :: {binary(), integer()}, column :: {binary(), integer()}) :: VegaLite.t()where both
rowandcolumnare tuples containing the index and field of the current and row and column respectively.You are free to specify any function for every cell of the grid.
Styling Options
:height(integer/0) - Height of the image:title(String.t/0) - The title of the graph:width(integer/0) - Width of the image
Notice that if set width and height will be applied to individual sub plots. On
the other hand title is applied to the composite plot.
Examples
By default a scatter plot will be drawn for all pairwise plots:
fields = ["petal_width", "petal_length", "sepal_width", "sepal_length"]
Tucan.pairplot(:iris, fields, width: 130, height: 130)You can color the points by another field in to add some semantic mapping. Notice
that you need the recursive option to true for the grouping to be applied on all
internal subplots.
fields = ["petal_width", "petal_length", "sepal_width", "sepal_length"]
Tucan.pairplot(:iris, fields, width: 130, height: 130)
|> Tucan.color_by("species", recursive: true)By specifying the :diagonal option you can change the default plot for the diagonal
elements to a histogram:
fields = ["petal_width", "petal_length", "sepal_width", "sepal_length"]
Tucan.pairplot(:iris, fields, width: 130, height: 130, diagonal: :histogram)
|> Tucan.color_by("species", recursive: true)Additionally you have the option to configure a plot_fn with which we can go crazy and
modify any part of the grid based on our needs. plot_fn should accept as input a VegaLite
struct and two tuples containing the row and column fields and indexes. In the following
example we draw differently the diagonal, the lower and the upper grid. Notice that we don't
call color_by/3 since we color differently the plots based on their index positions.
Tucan.pairplot(:iris, ["petal_width", "petal_length", "sepal_width", "sepal_length"],
width: 150,
height: 150,
plot_fn: fn vl, {row_field, row_index}, {col_field, col_index} ->
cond do
# For the first two diagonal elements we plot a histogram, no
row_index == col_index and row_index < 2 ->
Tucan.histogram(vl, row_field)
row_index == 2 and col_index == 2 ->
Tucan.stripplot(vl, row_field, group: "species", style: :tick)
|> Tucan.color_by("species")
|> Tucan.Axes.put_options(:y, labels: false)
# For the other diagonal plots we plot a histogram colored_by the species
row_index == col_index ->
Tucan.histogram(vl, row_field, color_by: "species")
# For the upper part of the diagram we apply a scatter plot
row_index < col_index ->
Tucan.scatter(vl, col_field, row_field)
|> Tucan.color_by("species")
# for anything else scatter plot with a quantitative color scale
# and size
true ->
Tucan.scatter(vl, col_field, row_field)
|> Tucan.size_by("petal_width", type: :quantitative)
end
end
)Grouping
@spec color_by(vl :: VegaLite.t(), field :: binary(), opts :: keyword()) :: VegaLite.t()
Adds a color encoding for the given field.
Options
:recursive(boolean/0) - If set the grouping function will be applied recursively in all valid sub plots. This includes both layers and concatenated plots. The default value isfalse.
opts can also contain an arbitrary set of vega-lite supported options that
will be passed to the underlying encoding.
@spec facet_by( vl :: VegaLite.t(), faceting_mode :: :row | :column, field :: binary(), opts :: keyword() ) :: VegaLite.t()
Apply facetting on the input plot vl by the given field.
This will create multiple plots either horizontally (:column faceting mode),
vertically (:row faceting mode) or arbitrarily (:wrapped mode). One plot will
be created for each distinct value of the given field, which must be a
categorical variable.
In the case of :wrapped a :columns option should also be provided which
will determine the number of columns of the composite plot.
opts is an arbitrary keyword list that will be passed to the :row or :column
encoding.
Facet plots
Facet plots, also known as trellis plots or small multiples, are figures made up of multiple subplots which have the same set of axes, where each subplot shows a subset of the data.
Examples
Tucan.scatter(:iris, "petal_width", "petal_length")
|> Tucan.facet_by(:column, "species")
|> Tucan.color_by("species")With :wrapped mode and custom sorting:
Tucan.density(:movies, "IMDB Rating", color_by: "Major Genre")
|> Tucan.facet_by(:wrapped, "Major Genre", columns: 4, sort: [op: :mean, field: "IMDB Rating"])
|> Tucan.Legend.set_enabled(:color, false)
|> Tucan.set_title("Density of IMDB rating by Genre", offset: 20)@spec fill_by(vl :: VegaLite.t(), field :: binary(), opts :: keyword()) :: VegaLite.t()
Adds a fill encoding for the given field.
Options
:recursive(boolean/0) - If set the grouping function will be applied recursively in all valid sub plots. This includes both layers and concatenated plots. The default value isfalse.
opts can also contain an arbitrary set of vega-lite supported options that
will be passed to the underlying encoding.
@spec shape_by(vl :: VegaLite.t(), field :: binary(), opts :: keyword()) :: VegaLite.t()
Adds a shape encoding for the given field.
Options
:recursive(boolean/0) - If set the grouping function will be applied recursively in all valid sub plots. This includes both layers and concatenated plots. The default value isfalse.
opts can also contain an arbitrary set of vega-lite supported options that
will be passed to the underlying encoding.
@spec size_by(vl :: VegaLite.t(), field :: binary(), opts :: keyword()) :: VegaLite.t()
Adds a size encoding for the given field.
By default the type of the field is set to :quantitative. You can override it in the
opts by setting another :type.
Options
:recursive(boolean/0) - If set the grouping function will be applied recursively in all valid sub plots. This includes both layers and concatenated plots. The default value isfalse.
opts can also contain an arbitrary set of vega-lite supported options that
will be passed to the underlying encoding.
@spec stroke_dash_by(vl :: VegaLite.t(), field :: binary(), opts :: keyword()) :: VegaLite.t()
Adds a stroke_dash encoding for the given field.
Options
:recursive(boolean/0) - If set the grouping function will be applied recursively in all valid sub plots. This includes both layers and concatenated plots. The default value isfalse.
opts can also contain an arbitrary set of vega-lite supported options that
will be passed to the underlying encoding.
Utilities
@spec concat(vl :: VegaLite.t(), plots :: [VegaLite.t()]) :: VegaLite.t()
Concatenates the given plots.
This corresponds to the general concatenation of vega-lite (wrappable).
@spec flip_axes(vl :: VegaLite.t()) :: VegaLite.t()
Flips the axes of the provided chart.
This works for both one dimensional and two dimensional charts. All positional channels that are defined will be flipped.
This is used internally by plots that support setting orientation.
@spec hconcat(vl :: VegaLite.t(), plots :: [VegaLite.t()]) :: VegaLite.t()
Concatenates horizontally the given plots.
@spec hruler(vl :: VegaLite.t(), position :: number() | binary(), opts :: keyword()) :: VegaLite.t()
Adds a horizontal line at the given h position.
For supported options check line/4.
@spec layers(vl :: VegaLite.t(), plots :: [VegaLite.t()]) :: VegaLite.t()
Creates a layered plot.
This is a simple wrapper around VegaLite.layers/2 which by default adds
the layers under an empty plot.
@spec new(plotdata :: plotdata(), opts :: keyword()) :: VegaLite.t()
Creates if needed a VegaLite plot and adds data to it.
The behaviour of this function depends on the type of plotdata:
- if a
VegaLite.t()struct is passed then it is returned unchanged. - If it is a binary it is considered a url and the
VegaLite.data_from_url/2is called on a newly createdVegaLitestruct. - if it is an atom then it is considered a
Tucan.Datasetand it is translated to the dataset's url. If the dataset name is invalid an exception is raised. - in any other case it is considered a set of data values and the values are set
as data to a newly created
VegaLitestruct.
@spec ruler( vl :: VegaLite.t(), axis :: :x | :y, position :: number() | binary(), opts :: keyword() ) :: VegaLite.t()
Adds a vertical or horizontal ruler at the given position.
position can either be a number representing a coordinate of the x/y-axis or a
binary representing a field. In the latter case an aggregation can also
be provided which will be used for aggregating the field distribution
to a single number. If not set defaults to :mean.
axis specifies the orientation of the line. Use :x for a vertical
line and :y for a horizontal one.
Options
:aggregate(atom/0) - The aggregate to used for calculating the line's coordinate The default value is:mean.
Data Grouping Options
:color_by(String.t/0) - If set a data field that will be used for coloring the data. It is considered:nominalby default.
Styling Options
:line_color(String.t/0) - The color of the line The default value is"black".:stroke_width(integer/0) - The stroke width in pixels The default value is1.
Encodings Custom Options
All Tucan plots are building a VegaLite specification based on some sane
default parameters. Through these encodings options you are free to set any
vega-lite supported option to any encoding channel of the plot.
Notice that if set they will be merged with any option set by the plot. Since they have a higher precedence they may override the default settings and affect the generated plot.
You can set an arbitrary keyword list. Notice that the contents are not validated.
:color(keyword/0) - Extra vega lite options for the:colorencoding. The default value is[].
Examples
You can add a vertical ruler to any x-axis point:
Tucan.scatter(:iris, "petal_width", "petal_length")
|> Tucan.ruler(:x, 1.1, stroke_width: 3, line_color: "blue")
|> Tucan.ruler(:x, 1.4, line_color: "green")Additionally you can can add a vertical line to an aggregated value of a data field. For example:
Tucan.scatter(:iris, "petal_width", "petal_length")
|> Tucan.ruler(:x, "petal_width", line_color: "red")You can add multiple lines for each group of the data if you pass the
color_by option. Also you can combine vertical with horizontal
lines.
Tucan.scatter(:iris, "petal_width", "petal_length", color_by: "species")
|> Tucan.ruler(:x, "petal_width", color_by: "species", stroke_width: 3)
|> Tucan.ruler(:y, "petal_length", color_by: "species")@spec vconcat(vl :: VegaLite.t(), plots :: [VegaLite.t()]) :: VegaLite.t()
Concatenates vertically the given plots.
@spec vruler(vl :: VegaLite.t(), position :: number() | binary(), opts :: keyword()) :: VegaLite.t()
Adds a vertical line at the given x position.
For supported options check line/4.
Styling
@spec set_height(vl :: VegaLite.t(), height :: pos_integer()) :: VegaLite.t()
Sets the height of the plot (in pixels).
@spec set_size(vl :: VegaLite.t(), width :: pos_integer(), height :: pos_integer()) :: VegaLite.t()
Sets the plot size.
This sets both width and height at once.
@spec set_theme(vl :: VegaLite.t(), theme :: atom()) :: VegaLite.t()
Sets the plot's theme.
Check Tucan.Themes for more details on theming.
@spec set_title(vl :: VegaLite.t(), title :: binary(), opts :: keyword()) :: VegaLite.t()
Sets the title of the plot.
You can optionally pass any title option supported by vega-lite to customize the style of it.
Examples
Tucan.scatter(:iris, "petal_width", "petal_length")
|> Tucan.set_title("My awesome plot",
color: "red",
subtitle: "with a subtitle",
subtitle_color: "green",
anchor: "start"
)@spec set_width(vl :: VegaLite.t(), width :: pos_integer()) :: VegaLite.t()
Sets the width of the plot (in pixels).