✳flâneur — a map of the web's best reading

sbensu: Math intuitions on variance

blog.sbensu.com · 1,011 words · saved by 1 readers

This is a supplement to High Variance Management, where I build some intuition on the different probability distributions involved. Please read that post before reading this one. We can develop some intuitions for these concepts using a simple probability model of Broadway and Hollywood. I'll use Julia to simulate distributions and plot them: Let's assume that the quality of a given take from an actor follows a normal distribution with zero mean and variance of one, Normal(0, 1). This is a way of modeling the following intuitions: In Broadway, each take is the scene. So, how a scene goes looks like directly sampling from a normal distribution: In Hollywood, we can do two things differently: Here is how the new Hollywood distribution compares with the Broadway one: According to this toy model, Hollywood's resulting product can be much better: You can only do this when you lean into the variance. And you can only do that if there is no cost to bad takes. The Hollywood distribution is sim

sbensu: Math intuitions on variance Math intuitions on variance January 2023 This is a supplement to High Variance Management , where I build some intuition on the different probability distributions involved. Please read that post before reading this one. The normal and Gumbel probability distributions We can develop some intuitions for these concepts using a simple probability model of Broadway and Hollywood. I'll use Julia to simulate distributions and plot them: using Random, Distributions, Plots plotly() Random.seed!(123) Let's assume that the quality of a given take from an actor follows

Explore this link on the map →

related reading