PHASE 00% · GROW
scroll ↓   scrub ↑↓ to re-read

Portfolio · 2026

Kip Jordan

Data Science · Business Analytics · Visualisation

Story-led analytics, designed to land with leadership. Models, dashboards and AI implementations, all built to leave decisions on the table, not metrics to reconcile.

See the work Skip to summary ↓

What this is

Watch these foundational algorithms think.

Some of the core of data science, statistics and machine learning. We're running them on well-known, publicly available datasets to experiment. Scroll, and watch how it unfolds.

+ +

About

Working from Melbourne, at the seam between data, design and the decisions they support.

Story-led analytics is the through-line. Models that inform a decision, dashboards that leaders actually use, AI implementations placed where they shorten the path from question to answer.

+ +
01 · Personal Hermes Agent Workforce A small operational AI platform I actually depend on Hermes Agent · Python · Raspberry Pi 02 · Professional Workforce Intelligence From “what happened” to “what should happen next” Python · pandas · YAML 03 · University Streaming Recommendation Engine Lift confirmed in Group B engagement (p < 0.001) R · A/B test · regression View all projects ↗

Selected work

One from each stream.

Professional · Personal · Tools · University

+ +

Get in touch

For collaboration, questions, or to say hello.

Send a message and I’ll read it. I’m based in Melbourne, Australia, time zones permitting.

hello@kipjordan.com
+ +

Portfolio · 2026

Kip Jordan

Data Science · Business Analytics · Visualisation

Story-led analytics, designed to land with leadership. Models, dashboards and AI implementations, all built to leave decisions on the table, not metrics to reconcile.

A decision tree, on the Palmer Penguins

The Palmer Penguins are a modern machine-learning benchmark: 344 penguins of three species, measured across three islands off Antarctica between 2007 and 2009. Could you tell the species apart? A decision tree learns to, from a few body measurements alone, one yes-or-no question at a time. A handful of questions, one answer, and it sorts the species correctly about 97% of the time on penguins it had never seen.

A Galton board

A Victorian contraption that proves a modern point. At every peg, each ball goes left or right at random; nothing remembers the last. Hundreds of them, every drop a blind coin toss, and still a shape appears: the bell curve, the normal distribution behind averages, polls and error bars.

Finding hidden groups (DBSCAN)

Finding the groups hiding in data that hasn't been labelled. DBSCAN asks one thing of every point: how many neighbours are close by? Dense points chain into shapes a straight line could never separate, and the loners stay loners, flagged as noise, never forced to belong.

Sorting into groups (k-means)

Another way to cluster: drop a few centres, let every point pick its nearest, and repeat until it settles. A field of points waiting to be grouped; three centres drift, each claiming whatever's nearest; they settle, and every point belongs somewhere.

Explore

hello@kipjordan.com · Melbourne, Australia

↑ Return to the top