Georgios Giasemidis' homepage

Mid-range analysis of Euroleague - Part II

And its impact on teams' performance, player stats and a bonus

Posted on April 13, 2024

In a previous blogspot, Mid-range analysis of Euroleague - And its impact on teams’ performance, after the first half of the Euroleague season, I presented an analysis of mid-range shots and its impact on team performance and players stats. Here, I repeat the analysis with the full data of the... [Read More]

Tags: basketball euroleague analytics midrange

Mid-range analysis of Euroleague

And its impact on teams' performance

Posted on December 31, 2023

The play style of basketball has evolved over the past two decades. Mid-range shots have drastically decreased in favour of more efficient three-point attempts in the NBA, an excellent overview of the topic is presented in the book Sprawlball: A Visual Tour of the New Era of the NBA by... [Read More]

Tags: basketball euroleague analytics midrange

Chat with and visualise your data - Part I

Using the ChatGPT API, Streamlit and clever prompt engineering

Posted on September 30, 2023

At a recent data science sharing session at Choreograph, I delivered a presentation on developing a chatbot using ChatGPT and Streamlit. While this project draws inspiration from previous work, it introduces two pivotal enhancements: [Read More]

Tags: chatgpt api chatbox streamlit plot plotly prompt engineering

Chat with and visualise your data - Part II

Using the ChatGPT API, LangChain agents, Streamlit and clever prompt engineering

Posted on September 30, 2023

In the first part of this series of articles, I described how one can create interactive plots using data returned from ChatGPT. I overlooked the validity of the returned data and focused on demonstrating how a simple prompt can give us powerful data visualisations. [Read More]

Tags: chatgpt api chatbox streamlit plot plotly prompt engineering langchain agent

Build a chat-bot assistant expert

Using the ChatGPT API and Streamlit

Posted on May 29, 2023

In the previous post, I described how to build an expert assistant, which answers only questions in a specific field, using ChatGPT API, streamlit and appropriate prompt engineering. However, the previous version of the app lacked a chat-box functionality. [Read More]

Tags: chatgpt api chatbox streamlit app basketball marketing advertising

How to make your own app of an expert assistant

Using the ChatGPT API and Streamlit

Posted on May 8, 2023

I was playing around with the ChatGPT API and thought what a better way to test it by making my own specialised assistant, which answers only questions on a specific topic/field/subject/industry. In the meantime, I was reading about prompt engineering for generative AI applications, like ChatGPT, see this free online... [Read More]

Tags: chatgpt api streamlit app basketball marketing advertising

Errors in Euroleague shot location data

Posted on November 6, 2022

A recent controversy broke out in the Euroleague game of Real Madrid against Virtus Bologna, where an extra point was suddenly added to Real Madrid’s score, see here, followed by errors in the game statistics, see also here. These errors reminded me of errors and inconsistencies I had encountered in... [Read More]

Tags: euroleague basketball analytics data shooting data

Good and Bad Practices and Habits in Data Science

A discussion at the WPP Open Data & AI "Speaker's Corner" Series

Posted on May 21, 2022

As a member of the WPP Open Data & AI community, I was asked to facilitate a discussion for the new series of sessions, “the Speaker’s corner”. I happily accepted the invitation, the community is very enthusiastic with loads of technical and domain expertise. However, I wasn’t sure what I... [Read More]

Tags: data science ai datascience bestpractices habits

An ABM presentation at WPP Open

An artistic visualisation of an ABM talk

Posted on April 20, 2022

I was invited to give a presentation at the WPP Open Data & AI session about a recent project I worked for (WPP is the parent company of Choreograph, the company I work for). [Read More]

Tags: abm agentbasedmodelling ai data covid visualisation artwork

Lessons from a review of low-voltage load forecasting

Challenges and Recommendations

Posted on November 24, 2021

Many projects and new hobbies started during the lockdowns over the past year and half. I was lucky enough to be evolved in a project with my very good colleagues, Stephen Haben, Siddharth Arora, Danica Vukadinović Greetham and Marcus Voss. (preprint version on arxiv.org and published version on Applied Energy)... [Read More]

Tags: forecasting forecast low-voltage load demand energy electricity review