Feed of long-form posts of MPN. Blog about technology, statistics, and other fun stuff.
I wanted to learn more about MCP, so I've built an MCP server that adds regression analysis capabilities to AI models. Here, we'll discuss what is MCP, why I chose linear regression, how to test an MCP server, and some limitations I currently see regarding the protocol and its implementations.
How I built yet another AI docstring generator.
In this post, we'll explore the mathematical foundations and implementation of bagging, a popular strategy to reduce model variance, and random forests, an algorithm that uses decision trees and bagging.
Implementing classification and regression trees (CART) in Python with categorical feature support.
This is the first in a series of posts about decision trees in the context of machine learning. The goal here is to provide a foundational understanding of decision trees and to implement them.
Exploring regression to the mean
Exploring the differences between lists in Go and Elixir
Tutorial de como criptografar um disco com LUKS, TPM + senha no Linux
Tutorial de como transformar seu computador antigo em um servidor com Apache Spark e Apache Hadoop
Last time, we used decision trees, binarization and logistic regression to predict heart failure mortality in a public dataset. Here, we'll use Cox regression to more adequately model survival data.
We'll predict heart failure mortality using a simplified model.
O que exatamente é a eficácia de uma vacina? Esse número se tornou muito popular, mas nem todas as interpretações sobre ele estão corretas.
What exactly is a p-value? Let's define the p-value and then look at what it is not.
Nós devemos focar em explicar por que terapias ineficazes podem parecer serem muito eficazes e mostrar os benefícios de ensaios clínicos de qualidade.
We should focus on explaining why ineffective therapies can appear to be very effective and show the benefits of proper clinical trials.
This post is intended to explain the basics of outlier detection and removal and, more specifically, to highlight some common mistakes.
We'll use Recurrent Neural Networks to classify the Sentiment140 dataset into positive or negative tweets.
A sentiment classification project utilizing logistic regression on the Sentiment140 dataset.
Exploratory data analysis is essential to construct hypothesis. Today we’ll explore the WHO Suicide Statistics database (version from Kaggle). It consists of a single CSV table, with 43776 instances of merely 6 variables.
Let's use PyTorch to analyze 3D fractals by implementing the box-counting algorithm and calculate the fractal dimension and lacunarity of a given image.
Implementing an algorithm for playing Battleship that uses heuristics.