dataoutsider 1.2.3

Data visualization toolbox.

Welcome to the dataoutsider visualization toolbox!

Featuring:

• Multi-Chord Diagram
  • 
• Pie-Tree Chart (or Pie-Map Chart)
  • 
• Bubble Series Chart
  • 

Multi-Chord Diagram

Have too many sets for a Venn Diagram?

1. Prepare groups a. Select 2 fields that identify 1. a group (ex: sports event) and 2. an entity (ex: athlete) to identify the many-to-many relationships (ex: many athletes to many events) or b. Pre-group data and provide each unique group combination (ex: athletes that competed in only the [100 meter] vs those that competed in only the [100 meter & 200 meter], etc.) and associated entity count (ex: # of athletes per group)
2. Configure the parent-group order
3. The multi-chord diagram will create bonds between many-to-many relationships, sized by the entity count per each unique group combination (ex: 10 athletes competed in only the [50 meter, 100 meter, & meter], 23 athletes competed in only the [100 meter], etc.)

Assumes 1 row = 1 unit for 1. a.

Example 1 (1. a.):

from dataoutsider import multi_chord as mc

df = mc.load_olympics_df()
df = df.loc[(df['Year'] >= 2014) & (df['Sex'] == 'F') & (df['Sport'] == 'Speed Skating')]
outer = 'Event'
inner = 'ID'
lookup = mc.multi_chord_get_alias(df, outer, inner)
print(lookup)
df_mc = mc.multi_chord_alias(df, outer, inner, percent = 75.)

df_mc_venn = mc.multi_chord_venn(df_mc)

mc.multi_chord_plot(df_mc, level = 4, transparency = 0.5)

Example 2 (1. b.):

from dataoutsider import multi_chord as mc
import pandas as pd

data = [['a', 25.5], ['a,b', 15], ['a,c', 14.4], ['a,c,d', 5], ['c,d', 13], ['d', 14], ['c,b', 10], ['b', 7]]
df = pd.DataFrame(data, columns = ['group', 'value'])
df_mc = mc.multi_chord_on_groups_alias(df, percent=33.)
df_mc_venn = mc.multi_chord_venn(df_mc)
mc.multi_chord_plot(df_mc, level = 3, transparency = 0.5)

Example 3 (1. a.):

from dataoutsider import multi_chord as mc
import pandas as pd

data = [['5', 'a'], ['5', 'b'], ['5', 'd'], ['2', 'b'], ['2', 'c'], ['3', 'b'], ['3', 'd'], ['4', 'c']]
df = pd.DataFrame(data, columns = ['inner', 'outer'])
lookup = mc.multi_chord_get_alias(df, 'outer', 'inner')
print(lookup) # use outer_ID (alias) to set manual order
df_mc = mc.multi_chord_alias(df, outer='outer', inner='inner', order='1,3,2,4', percent=33.) #, order = 'b,c,d,a'
mc.multi_chord_plot(df_mc, level = 3, transparency = 0.5)

Example 4 (1. b. version of Example 3):

from dataoutsider import multi_chord as mc
import pandas as pd

data = [['a,b,d', 1], ['b,c', 1], ['b,d', 1], ['c', 1]]
df = pd.DataFrame(data, columns = ['group', 'value'])
df_mc = mc.multi_chord_on_groups_alias(df, order = 'b,c,d,a', percent=33.)
mc.multi_chord_plot(df_mc, level = 3, transparency = 0.5)

Functions:

multi_chord_alias(df, outer, inner, percent=100., order=None, buffer = 1., elements_offset = 0.04, elements_height = 0.04, group_offset = 0., group_height = 0.04)

multi_chord_get_alias(df, outer, inner)

multi_chord_on_groups_alias(df_chord, percent=100., order=None, buffer = 1., elements_offset = 0.04, elements_height = 0.04, group_offset = 0., group_height = 0.04)

multi_chord_venn(df_mc)

multi_chord_plot(df_mc, level = None, transparency = 0.5)

• multi_chord_alias: Multi-Chord Diagram generator that encodes the supplied groups with an alias, output: Pandas DataFrame
• multi_chord_get_alias: retreive the alias for the supplied groups (can use to create a manual ordering)
• multi_chord_on_groups_alias: Multi-Chord Diagram generator based on pre-defined groupings, output: Pandas DataFrame
• multi_chord_venn: output for creating an Upset plot, output: Pandas DataFrame
• multi_chord_plot: generate a plot using Matplotlib

Parameters:

Parameter	Values	Description
df	(DataFrame)	Pandas DataFrame
outer	(string)	Column name of group
inner	(string)	Column name of group entity
percent	(float: 0-100)	Percent of full circle to draw within
order	(csv string)	Group order
buffer	(float)	Radial distance between groups
elements_offset	(float)	Distance offset of element legend for entities
elements_height	(float)	Height of element legend for entities
group_offset	(float)	Distance offset of group legend
group_height	(float)	Height of group legend
df_mc	(DataFrame)	Pandas DataFrame of multi-chord results
level	(int)	Bold chords by count of groups in chord (optional)
transparency	(float: 0-1)	Add color transparency (optional)

Pie-Tree Chart

Need to display hierarchical density as pie-shaped areas?

1. Select [1 to n] categorical columns from a dataframe to group by
2. Configure the chart ordering and hierarchy orientation
3. The pie-tree chart will create a hierarchical set of areas sized by the number of rows in each group at each level

Assumes 1 row = 1 unit

Example:

from dataoutsider import pie_tree as pt

df = pt.load_aircraft_df()
levels = ['Registrant', 'Aircraft', 'Engine', 'seats_bin']

inner_radius = 0.5
outer_radius = 2.0
starting_angle = 0.0
ending_angle = 360.0
point_resolution = 200
pie_tree_df = pt.pie_tree_calc(
    df, levels, 
    inner_radius, 
    outer_radius, 
    starting_angle, 
    ending_angle, 
    point_resolution)

pt.pie_tree_plot(pie_tree_df, 4)

Functions:

def pie_tree_calc(df, groupers, r1, r2, start_angle, end_angle, points, default_sort = False, default_sort_override = True, default_sort_override_reversed = False, all_vertical = False)

pie_tree_plot(pie_tree_df, level = 1, transparency = 0.5, line_level = 0)

• pie_tree_calc: Pie-Tree (Pie-Map) Chart generator, output: Pandas DataFrame
• multi_chord_plot: generate a plot using Matplotlib

Parameters:

Parameter	Values	Description
df	(DataFrame)	Pandas DataFrame
groupers	list of (string)	Columns names in df
r1	(float)	Innder radius
r2	(float)	Outer radius (> inner radius)
start_angle	(float)	Angle to start drawing
end_angle	(float)	Angle to end drawing
points	(int)	Resolution for curve drawing
default_sort	True/False	Default: False, True: pandas sort, False: data sort
default_sort_override	True/False	Default: True, True: overrides default_sort
default_sort_override_reversed	True/False	Default: False, sort areas True: desc, False: asc
all_vertical	True/False	Default: False, True: break levels vertically, False: alternate
level	(int)	Hierarchy level to plot (optional)
transparency	(float: 0-1)	Add color transparency (optional)
line_level	(int)	Hierarchy level to bold (optional)

Variations:

Bubble Series Chart

Need to visualize sized bubbles that have a positional feature?

1. Identify the columns names in the data for item, position, and size (circle area)
2. Add an optional radius buffer to separate the circles
3. The bubble series chart will pack all of the sized circles deterministically according to their positions (along the horizontal axis)

Assumes 1 row = 1 unit

Example:

from dataoutsider import bubble_series as bs

data = [
    ('a',-5.2,4.1),
    ('b',-3.2,3.5),
    ('c',2,2),
    ('d',1,7),
    ('e',5,5),
    ('f',13,4),
    ('g',12,3),
    ('h',1.5,0.5)]
df = pd.DataFrame(data, columns = ['name', 'position', 'size'])

df_bs = bs.bubble_series_calc(df, 'name', 'position', 'size', 0.0)
bs.bubble_series_plot(df_bs)

Functions:

bubble_series_calc(df, item_col_name, position_col_name, size_col_name, r_buffer=0.0)

bubble_series_plot(bubble_series_calc_df, transparency = 0.5)

• bubble_series_calc: Bubble Series Chart generator, output: Pandas DataFrame
• bubble_series_plot: generate a plot using Matplotlib

Parameters:

Parameter	Values	Description
df	(DataFrame)	Pandas DataFrame
item_col_name	(string)	Columns name of items
position_col_name	(string)	Columns name of item's positions
size_col_name	(string)	Columns name of item's sizes
r_buffer	(float)	radius buffer to add separation between bubbles (optional)
transparency	(float: 0-1)	Add color transparency (optional)

Tableau users

Output:

• Polygon chart: Columns: [x], Rows: [y], Path: [path], Detail: varies by algorithm, use grouping fields as needed (see examples below)

Examples:

• Multi-Chord Diagram: Sink or Swim
• Pie-Tree Chart: Takeoff
• Bubble Series Chart: Black Hole Sun

License

The dataoutsider package is dual-licensed:

• For non-commercial use, dataoutsider is available under the Non-Commercial Use License (LICENSE-NC). See the LICENSE-NC file for more details. This includes specific provisions against unauthorized adaptations of the software's core functionality or algorithms in any other programming language or environment without explicit permission.

• For commercial use, a separate Commercial Use License (LICENSE-COM) is required. This license includes provisions against the creation and commercial exploitation of adaptations or derivative works without obtaining a specific commercial license. Please contact Nick Gerend at [e-mail] for terms and inquiries.