Introducing the Crystal Bar Chart: Visualizing Sequential Differential Clustering | by Nick Gerend | Dec, 2023


Thank you for reading this post, don't forget to subscribe!

Find out about Crystal Bar Charts and create your individual with Python

Nick Gerend

Towards Data Science
Crystal Bar Chart by Nick Gerend

There are various methods to higher perceive a collection of knowledge. Whether or not or not it’s a set of temperatures, take a look at scores, grocery costs, step counts or UFO sightings, there’s at all times one thing to study in regards to the nature of one thing by the lens of knowledge evaluation and visualization.

Within the following, I’ll describe a brand new approach for visualizing a collection of knowledge that makes use of easy threshold-based clustering. I hope you’ll discover it to be a contemporary method to pair with typical strategies of visualization like histograms and field plots for gathering a brand new perspective on a characteristic of curiosity.

Numerically

How does one set itself aside? By way of a single numerical attribute, let’s assume a easy threshold is the usual by which ordered values are in comparison with each other and separated into teams. Take for instance the next sequence:

[ 0 , 1 , 1 , 2 , 3 , 5 , 8 , 13 , 21 , 34 , 55 , 89 , 144 ]

That is after all the primary 13 digits of the Fibonacci sequence. If we take the distinction of every worth and the prior worth within the sequence, beginning with third place, we get the identical sequence shifted ahead by 2 steps:

[ _ , _ , 0 , 1 , 1 , 2 , 3 , 5 , 8 , 13 , 21 , 34 , 55 ]

Now let’s set a threshold of 5 and assume something larger than 5 delineates a brand new cluster as we traverse the sequence from left to proper. This may consequence within the first 8 gadgets being grouped collectively (values of 0–13) with every subsequent merchandise being the one merchandise in its group since all remaining variations exceed 5.

For the Fibonacci sequence this methodology yields a predictable consequence, however in exploratory knowledge evaluation all sorts of patterns might emerge. I’ll describe discover these patterns with a brand new sort of knowledge visualization, beginning with a few acquainted chart sorts beneath for comparability.

With a Histogram



Leave a Reply

Your email address will not be published. Required fields are marked *