Data Science on LittleBiggler

Regularization

Mon, 29 Sep 2025 16:48:04 +0000

In hindsight, I was able to suss out a handful of insane outliers (like 25 sigma outliers) that were probably typos, and were most likely causing the regression line to whip back and forth like the infamous 2016 NYTimes election needle.

I thought I was dealing with multi-collinearity and simplex group problems only, but lurking in the data the entire time were some special surprises. It took me way too long to figure out what was going on I will never look at a box plot the same way again.

Bayes: Odds Vector Form

Thu, 14 Nov 2024 20:31:00 +0000

Regardless of whether you subscribe to the view of causality as the ur-metaphor, the OVF will help you learn Bayes in a way that prepares you for more sophisticated pattern recognition.

Specifically, I hope to show how the Odds Vector Form (hereinafter also referred to as the OVF) of Bayes Theorem yields better Bayes formulas than the most commonly used Probability Form. The visual elegance of the vector form commits itself better to memory. Stripped of clunky redunancies, OVF Bayes allows one to add many more variable dimensions easily (by hand, even). A relatively junior coder could write a classification formula that doesn't rely on a library, or even use pencil and paper to calculate the posterior probability of complex scenarios.

When Is Now? A Secular Proof of the Existence of God

Wed, 11 Sep 2024 18:24:00 +0000

At first glance, this post has nothing to do with data science. I hope that by the time you finish, you will understand why it has everything to do with data science. The connection is the distinction between objective reality (being an object) vs. being a trace of an object.

Your Own Personal Loss Function

Wed, 29 May 2024 17:30:00 +0000

For me, heaven is a place where I can take every mathematics class that Professor Gilbert Strange ever taught. Most of the time, I find the pedestal narratives applied to professors at big research universities to be farcical. Why? In graduate school at UChicago, I found many of the tenured professors (who shall remain unnamed of course) to be pompous and not only uninterested in teaching, but actually contemptuous of it. Not Gil Strang, though. Gil Strang wanted to teach, and did it well. He deserves to be lionized, misogynistic jokes about Monica Lewinsky notwithstanding (it's a fallen world, after all). Last year, Professor Strang gave his final lecture. I expect he will not even know what a legend he truly is until Judgment Day.

Baking With Math: Bootstrapping

Fri, 01 Mar 2024 19:54:40 +0000

Bootstrapping is an option you take when the data in hand is kind of shitty, so you shuffle it, and resample it to clone more data, and pray that the clone represents reality (the population).

Standard deviation, if you recall from our Baking With Math series, is a measure of how well mixed your batter, which represents, portentously, reality, is. If you take a bunch of spoonful's of your batter, in other words, the mean mixed-upness of those spoonfuls is fairly likely to represent how mixed up the data is in the population. You could get one spoonful that is mostly egg vs not-egg, and get one spoonful that is mostly sugar vs not-sugar, and still your mixed-upness gauge could still be accurate with respect to the overall population.

Counter-Factual Universe

Mon, 29 Jan 2024 15:31:17 +0000

"The Null Hypothesis" is a theoretical alternate universe, used to measure statistical significance. The standard deviation and mean are quantitative measures, used to bootstrap or simulate the probability distribution of a given facet of your alternate universe.

If the piece of evidence you encountered, were it found your alternate universe, would be super super super super super super super duper unlikely, then your alternate universe probably isn't reality, which means your null hypothesis should probably be rejected.

A New Way to Visualize Bayes

Mon, 02 Oct 2023 19:23:00 +0000

A lot of instructional resources go into visualizing Bayes Theorem, to help the visual learners out there. This post is about how I was ultimately able to wrap my head around Bayes. I hope it is helpful to you!

You Wanna Bet You Can Derive Bayes?

First of all, as we proceed through this exercise, I want everyone to understand that Bayes Theorem is nothing special. Don't get me wrong!!! The Bayes formula is elegant because of its almost infantile simplicity. Bayes distills the quintessential formula by which humans learn about the world. And compute power unleashed by modern semiconductors has enabled machines to crunch countless parameters very quickly, so as to yield highly accurate predictions. None of this is unimpressive... But let's not lose sight, in spite of all of the hype, of how simple and mundane the core idea of Bayes' Theorem is.

Labor Force Participation and "Disability"

Wed, 20 Sep 2023 14:39:46 +0000

Data Source: Bureau of Labor Statistics - Labor Force Participation

In another post I explored the relationship between labor force participation for women, versus fertility rates. In that analysis, I stumbled upon the fact that women's labor force participation in the United States peaked around 2008, and has only declined since then. Initially, I thought this could be due to millennial women dropping out of the labor force to raise babies, since the fertility data shows that women in their thirties are having more babies now than they did in decades past.

Women's Labor Force Participation vs. Fertility

Wed, 30 Aug 2023 16:35:16 +0000

Data Sources: United Nations Population Division and US Bureau of Labor Statistics

In his excellent book, The Decadent Society, Ross Douthat observes that women tend to go on strike, when society expects them to have children and a career. And they go on strike, not by demanding better working conditions, which might be more conducive to having children. They go on strike by refusing to have children.

Big Data Public School Rankings

Thu, 29 Sep 2022 17:23:03 +0000

Who cares about "Good Schools"?

Do me a favor: Pull up a real estate listing online... Maybe it's the dream house you have saved on Zillow, or a starter apartment near your job. Maybe you were not previously aware that every single real estate listing displays public school rankings. Go ahead, find the school rating on that property listing. It is there, and if you are a parent, it's often the first thing you scroll down for.