ParallelPlots

General

This Project is for the TU-Berlin Course "Julia Programming for Machine Learning"
Please make sure, that Julia 1.10 is used!

This Module will return you a nice Scene you can use to display your Data with Parallel Coordinates

This Module was created with PkgTemplates.jl

Mathematical Background

Parallel coordinates transform high-dimensional data into a 2D visualization by aligning axes parallel to each other. Key transformations include:

1. Axis Shifting

Shifting the origin of a coordinate system from $O{a, b}$ to $O'{a', b'}$ transforms coordinates as:
$x' = x - a, \quad y' = y - b$

2. Normalization

To compare variables on different scales, data is normalized to $[0, 1]$:
$x_{\text{norm}} = \frac{x - \min(x)}{\max(x) - \min(x)}$

3. Rotation (for Orthogonal Systems)

For rotated axes by angle $\alpha$ in a Cartesian system:
$$ x' &= x \cos \alpha + y \sin \alpha \ y' &= -x \sin \alpha + y \cos \alpha \end{align*} $$

These principles underpin how parallel coordinates handle scaling and alignment of axes.

Getting Started

Install Dependencies & Use Package

Please refer to this Link for Installation of Julia

You need to use the package (1-3) and install the dependencies (4-5)

1. Open Julia with julia in your command prompt
2. Open the package manager with ]
3. Using our Package

   • activate /path/to/package
     or
     activate . when Julia was opened with command prompt already in package path

   • you will then see: (ParallelPlots) pkg>

4. go back to julia> by pressing CMD+C
5. Import ParallelPlots to download Dependencies and use the Package from Command Line

Usage

Available Parameter

Parameter	Default	Example	Description
normalize::Bool	false	normalize=true	If the Data should be normalized (min/max)
title::String	""	title="My Title"	The Title of The Figure,
colormap	:viridis	colormap=:thermal	The Colors of the Lines
color_feature	nothing	color_feature="weight"	The Color of the Lines will be based on the values of this selected feature. If nothing, the last feature will be used
feature_labels	nothing	feature_labels=["Weight","Age"]	Add your own Axis labels, just use the exact amount of labes as you have axis
feature_selection	nothing	feature_selection=["weight","age"]	Select, which features should be Displayed. If color_feature is not in this List, use the last one
curve	false	curve=true	Show the Lines Curved
show_color_legend	nothing	show_color_legend=true	Show the Color Legend. If parameter not set & color_feature not shown, it will be displayed automaticly

Examples

julia> using ParallelPlots
julia> parallelplot(DataFrame(height=160:180,weight=60:80,age=20:40))

# If you want to normalize the Data, you can add the value normalized=true, default is false
julia> parallelplot(DataFrame(height=160:180,weight=reverse(60:80),age=20:40),normalize=true)

# If you want to set the size of the plot (default width:800, height:600)
julia> parallelplot( DataFrame(height=160:180,weight=60:80,age=20:40), figure = (resolution = (300, 300),) )

# You can update as well the Graph with Observables
julia> df_observable = Observable(DataFrame(height=160:180,weight=60:80,age=20:40))
julia> fig, ax, sc = parallelplot(df_observable)

# If you want to add a Title for the Figure, sure you can!
julia> parallelplot(DataFrame(height=160:180,weight=reverse(60:80),age=20:40),title="My Title")

# If you want to specify the axis labels, make sure to use the same number of labels as you have axis!
julia> parallelplot(DataFrame(height=160:180,weight=reverse(60:80),age=20:40), feature_labels=["Height","Weight","Age"])

# Adjust Color and and feature
parallelplot(df,
		# You choose which axis/feature should be in charge for the coloring
        color_feature="weight",
        # you can as well select, which Axis should be shown
        feature_selection=["height","age","income"],
        # and label them as you like
        feature_labels=["Height","Age","Income"],
        # you can change the ColorMap (https://docs.makie.org/dev/explanations/colors)
        colormap=:thermal,
        # ...and can choose to display the color legend.
        # If this Attribute is not set,
        # it will only show the ColorBar, when the color feature is not in the selected feature
        show_color_legend = true
    )

Please read the Docs for further Information

Working on this Package / Cheatsheet

1. Using the Package

   • activate /path/to/package
     or
     activate . when Julia was opened with command prompt already in package path

   • you will then see: (ParallelPlots) pkg>

2. Running commands

   • Adding external Dependencies
     • add DepName
   • Run Tests to check if Package is still working as intended
     • test
   • Build
     • build
   • Precompile
     • precompile

Create Docs

• move to ./docs folder with command line
• run julia --project make.jl