Events

Presentations

John Haidoulis - Effective Experimental Design and Data Analysis in Biosciences

This talk will provide an overview of the key stages involved in designing experiments, analysing and visualising biological data. We’ll dive into hands-on examples from plant science, including essential tools, programming languages & libraries, and statistical methods that are typically used. We will then briefly discuss the emerging integration of machine learning in the biological research pipeline

Giancarlo Erra (Altaire) - From Prompts to Production: what businesses ask about AI

Businesses are committed to AI adoption—with 78% of global companies now utilising AI—but the transition from "Prompts to Production" is currently stalled by organisational barriers, chiefly a lack of skills (50%) and a lack of vision among managers (43%). The technical imperative is therefore to drive pragmatic change by prioritising workforce upskilling (planned by 85% of employers) and the automation of processes (73%), focusing on empowering non-technical users from basic prompt engineering toward integrated, iterative solutions.

Workshop

Ben Askew-Gale (Natwest) - Python Software Design Workshop

As your Python programs grow and become more complex, they can quickly become hard to maintain, extend and reason about. Software design patterns help keep this complexity under control, making your code more understable and facilitating the scaling to larger applications.

In this workshop, we will take some bad Python code and refactor it step by step. You will learn to spot primitive obsession, replace conditionals with the Strategy pattern, and use Dependency Injection to write testable code.

Presentations

Dr Gabriel Harris (NIQ Brandbank) - Modern Tools for Python Development: Code for Collaboration

Data science projects are often messy with inconsistent environments and dependencies, and individual coding preferences. These challenges undermine reproducibility, code quality, and ultimately slow down collaboration. In this talk I will go through some of my best Python development tools and show you how they can automate much of the boilerplate work and enforce consistency and best practices across your team. These include:

• uv to install and manage Python versions, virtual environments, and dependencies
• pyproject.toml to isolate dependency groups and add tools configurations
• Ruff to line and format code
• pre-commit to manage Git pre-commit hooks
• Bonus: dev containers

Details

Welcome back to PyData Norwich (Formerly Data Science Norfolk). After a bit of a hiatus we have returned under the PyData umbrella!!

This event is primarily a chance to meet up again and discuss what's been going on in Python/Data Science recently. However a number of people have raised the possibility of having a website related to the group to help with things like

• Getting started in both Python and Data Science and how to improve those skills.
• Advice and tips when looking for jobs.
• Knowing where to find useful resources.

The plan is therefore to have some informal discussions about what would be useful and crowdsource some suggestions of relevant content (mainly links to good resources) tailored to Norwich.

If you do not wish to contribute or feel you don't have the experience you are still very welcome to join and listen to the discussions, Late-comers are also welcome.

Afterwards we will go to the pub!

Details

An inpromptu gathering. No talks this time!

We will be meeting at the Playhouse Bar for a drink and to discuss all things Python/Data-related (and more).

Whether you are just starting out or an experienced developer you are welcome to come along. We will be there from 5.30pm onwards.

Presentations

1. Ben Askew-Gale (NatWest) - The Rustification of Python

How Rust is making the Python ecosystem better

2. Patrick Ogilvie (Socialfunnel) - Why A/B Testing sucks

Re-thinking User Journeys and Conversion Rates with column-oriented and graph databases

Presentations

1. Mike Joyner (Hartree Partners / Bricks & Logic) - Predicting house prices without machine learning

When data quality is low and transparency and explainability is key, why a statistical based approach rather than traditional machine learning methods leads to more effective results

2. Richard Bensley (MariaDB) - Map, filter, and reduce, then and now

A quick look at what was originally implemented and how they have changed.

Presentations

1. Matt Brookson (Artlist) - AI with extra fries

Order fast food with the help of OpenAI and function calling. A fun guide to leveraging the OpenAI Python API to make applications.

2. Isaac Bowers-Barnard (Aviva) - RL and Insurance Pricing

A brief primer on reinforcement learning and its applications on pricing retail insurance on price comparison sites.

Presentations

1. Mark Harvey (Maplecroft) - Generating reliable libraries (with uv and nox)

An overview of how to ensure reliable backwards compatible libraries with tooling such as uv and nox

2. Yogi Louise Bear (Freelance Machine Learning Engineer) - Using NLP, graph theory, and qualitative approaches to identify markers of radicalisation in online conversations.

This talk will summarise a recent machine learning project which sought to recognise the roles individuals play in online conversations that lead to either radicalisation or the incitation of violence at large-scale events such as protests. The intention is to understand how extreme online discord operates with a view to using the information to prevent future negative events.

Presentations

1. Kev Sturman (Kaarbontech) - Harnessing the Power of Infrastructure as Code: Building with AWS CDK and Python.

Explore the capabilities of AWS Cloud Development Kit (CDK) with Python, define and deploy cloud infrastructure using code.

2. Chris Ballard (Justified AI) - Few shot learning with SetFit

Training a machine learning model normally requires labelled data. In this talk, Chris will talk about how you can train a model when labelled data is scarce, discussing the different techniques that can be used in this situation. He will introduce an approach called SetFit and how it can be used to train models that can be efficiently deployed in production.

Presentations

1. Richard Bensley (Vettabase Ltd) - Extending Python using Golang.

What is Go, why is it useful how can it be integrated into your Python projects?

2. Chris Joyner (Artlist) - Migrating from Pandas to DuckDB/Polars.

What are the DuckDB and Polars libraries used for data analysis/manipulation and why might they be preferable to Pandas?