Hackathon - Summer 2023

Just finished my real "proper" #hackathon project at Mountain Warehouse.

I did a mini one a couple of years back looking at mesh networking with Raspberry Pi with the retail team lead at the time. But that was a more like individual research as the two of us were working on operational things in parallel. And on the other there had been time critical projects to finish such as helping launch the MW Marketplace so I didn't have a chance to join other sessions.

So this time, I made sure that I could dedicate time to joining a project. I also had a project idea to pitch.

Investigate techniques and visualisation of data to identify sudden changes and outliers.

The reason for the interest in outliers is that they often tell us where something is going very wrong, or very right. For example if you make a sale where someone spends a lot of money, it can be useful to know why they did that so you can reproduce it. Or looking at the weights of products (which we need to predict shipping costs), if one pair of boots came in at 20Kg or 20g then that would indicate incorrect data as we could investigate it and correct it.

My team investigated some different algorithms; Tukey's Fences, Z-Scores and Hampel's Identifier. And we put some of those into a simple Blazor app for visualisation using the Plotly.js and the Blazor wrapper for that Plotly.Blazor. As a technical exercise it was a success in that we successfully identifier outliers and visualised them. But it was also a success as a team event with much more collaboration than my previous hackathon and also more than many "team building" events I've been on.

The presentation went well with the demo app acting both as the narrative and the content. There were also some great projects from other teams. Tools for managing errored RabbitMQ messages, a prototype web store in NextJS, a feedback application, application metrics using OpenTelemetry, Predictive modelling using ML and of course an OpenAI project.

And there was great evidence of both creativity and collaboration from all teams. Lots of people left the presentation with ideas they could take back to their own applications and projects. And it was a big boost to team motivation. So all in all, I think it was a successful event.

A massive thanks to Phil Sharpe for co-coordinating and to Mountain Warehouse for allowing us time to develop our skills both technically and as a team.