blank

Among spies, hackers and tabloids, independent component analysis (ICA) immediately brings to mind the cocktail party problem, the notorious signal processing problem it solves.

A big part of data science revolves around structured data, data that can be neatly organized into tables.

In pop science, entropy is considered a measure of disorder: a system has high entropy when it is disordered (e.g., my college bedroom), and low entropy when it is ordered (e.g., a chocolate box).

Imagine we have a set of 3 variables (\(U\), \(X\), and \(Y\)), with one of them being upstream of the other two (\(X \leftarrow U \rightarrow Y\)): \[U \sim N(0, 1)\] \[X = U + \varepsilon_X\] \[Y = U + \varepsilon_Y\] \[\varepsilon_X, \varepsilon_Y \sim…

Picture this.

Python veterans will be familiar with pip, poetry, pyenv, conda and a few other tools for managing projects, packages and environments. uv’s goal is to replace them all while being blazingly fast.

SHAP values are a model-agnostic method to quantify the contribution of any given feature to a model’s prediction.

In many scientific applications, the goal is to discover which features are truly associated with an outcome.

Random walk A random walk (RW) is a stochastic, discrete process.

In this article I discuss matrices associated to graphs.