InData Science CollectivebyFrancisco Mandujano-Reyes·Oct 18Foundation Models for Time Series: What Are They Learning? How Can We Use Them?Forecasting has traditionally meant building a model for a particular time series. But foundation models change the question.A response icon2A response icon2
InTowards AIbyFrancisco Mandujano-Reyes·Aug 17LLM Watermarking Detection is a Statistical Problem: Hypothesis-TestingAt the time of writing this LLMs watermarking is becoming significantly important. All major language models, including Claude, Gemini…A response icon1A response icon1
InData Science CollectivebyFrancisco Mandujano-Reyes·Jul 1Retrieval-Augmented Generation (RAG): LLM + Memory?Large language models have a hard constraint: their knowledge is frozen at training time. If we ask GPT-4 about something that happened…
InData Science CollectivebyFrancisco Mandujano-Reyes·Jun 27Quick and Easy Guide to Fine-Tuning LLMsLarge language models like Mistral, Llama, and Phi-2 are impressive as they are, but they are trained for general porpuses. When we need a…
Francisco Mandujano-Reyes·May 3LLMs as Time Machines: Running Experiments on the PastIn The Dead Past, Isaac Asimov imagines a device called chronoscope that allows people to observe the past. At first, it seems like a…
InData Science CollectivebyFrancisco Mandujano-Reyes·Apr 4What is Granger Causality? (Intuition + Math)Granger Causality: From Time Series to Directed Acyclic Graphs
Francisco Mandujano-Reyes·Feb 1I Used Design of Experiments to Optimize My Morning Coffee ☕📊During my PhD, I spent most of my time thinking about stochastic models, assumptions, and inference. But one morning, while making coffee…
Francisco Mandujano-Reyes·Nov 17, 2025Prompt Examples for Data & Applied Scientists (and Why They Work)Over the last few years, I’ve used LLMs as a coding partner, debugging assistant, research “brainstorm buddy,” writing editor, travel…
Francisco Mandujano-Reyes·Feb 14, 2025Cluster Analysis of Premier League PlayersThis is another soccer analysis. In this one, I will create a simple cluster analysis of players. I’ll use Premier League 2015/2016 Season…