Blog

Pedro Tabacof’s writing about machine learning, statistics, experiments, and engineering.

Blog

Technical notes on machine learning, statistics, experimentation, and engineering.

How I lost 1000€ betting on CS:GO — Practice

Part 2 of a series of 2 posts where I explore how I lost 1000 euros betting on CS:GO with machine learning (ML). This post covers the actual implementation of the solution: CS:GO, feature engineering, modelling, validation, backtesting and lessons learned.

Jul 11, 2024
Pedro Tabacof

How I lost 1000€ betting on CS:GO — Theory

Part 1 of a series of 2 posts where I explore how I lost 1000 euros betting on CS:GO with machine learning (ML). This post covers the foundations of e-sports betting with ML: financial decision-making, expected profits of a bet, multiple bets with the Kelly criterion, probability calibration, and the winner’s curse.

Jan 4, 2024
Pedro Tabacof

Real-Time ML Models with Serverless AWS

Using AWS Lambda and API gateway
In this post, I will explore how to deploy a real-time machine learning model using AWS Lambda and API Gateway. I will go over the following points:
Jun 4, 2023
Pedro Tabacof

Name classification with ChatGPT

How does it compare to machine learning language models?
I explore the problem of name classification with ChatGPT and three machine learning models of increasing complexity: from logistic regression to FastAI LSTM to Hugging Face…
Mar 27, 2023
Pedro Tabacof

The Hierarchy of Machine Learning Needs

In 1943, Abraham Maslow created the hierarchy of human needs, ranging from basic physiological needs to abstract concepts like self-actualization. In this article, I propose…
Mar 15, 2023
Pedro Tabacof

NeurIPS 2018: A Data Scientist’s Perspective

Two weeks ago in Montreal, NeurIPS (formerly known as NIPS) took place, the world’s largest conference on machine learning and artificial intelligence. Major advancements in…
Dec 17, 2018
Pedro Tabacof

How (not) to forecast an election

We use a hierarchical Bayesian model to show how a simple pro-Trump bias has a huge effect on forecasting election results. See the discussion on HackerNews.
Nov 16, 2016
Pedro Tabacof
No matching items