Matt Kaye – posts

A Gentle Introduction to Docker

I need to run my code somewhere other than my machine. How do I do it?

R

data science

This post is part of a series called The Missing Semester of Your DS Education.

Dependency Management

You should be using poetry, or renv, or conda, or something similar

data science

R

python

This post is part of a series called The Missing Semester of Your DS Education.

Experiment Tracking and Model Versioning

Knowing what you’ve tried, and what’s in production

data science

This post is part of a series called The Missing Semester of Your DS Education.

Workflow Orchestration

A Beginner’s Guide to Shutting Down Your Machine at Night

data science

This post is part of a series called The Missing Semester of Your DS Education.

Pull Requests, Code Review, and The Art of Requesting Changes

R

data science

This post is part of a series called The Missing Semester of Your DS Education.

Unit Testing Analytics Code

How do you know your code actually works?

R

data science

This post is part of a series called The Missing Semester of Your DS Education.

Writing Internal Libraries for Analytics Work

Or packages, or modules, or whatever you wish to call them

R

data science

This post is part of a series called The Missing Semester of Your DS Education.

Balancing Classes in Classification Problems

And why it’s generally a bad idea

R

data science

In my last post I wrote about common classifications metrics and, especially, calibration.

Calibration and Evaluating Classification Models

Metrics for probability predictions

data science

There’s something of a complexity trajectory in evaluating classification models that…

Interpreting AUC-ROC

data science

R

AUC goes by many names: AUC, AUC-ROC, ROC-AUC, the area under the curve, and so on. It’s an extremely important metric for evaluating machine…

Exploring the Tail Behavior of ESPN’s Win Probability Model

data science

statistics

sports analytics

It’s College Football Playoff season, which means I’ve been watching a lot of games lately. And I find myself complaining pretty often about how badly calibrated I think…

Sequential Testing

data science

statistics

The last post proposed a solution to the multiple testing problem that often invalidates A/B test results test planning. The idea is to calculate the sample sizes you need…

Calling A/B Tests

data science

statistics

a/b testing

In the last post, I gave a bird’s eye level overview of the mechanics of running an A/B test. But at the end, we reached a problem: We had two conversion rates – 20% and 25%…

Running A/B Tests

data science

statistics

a/b testing

This is the second post in a series on A/B testing. In the last post, I gave a high-level…

A/B Testing: A Primer

data science

statistics

a/b testing

This is the first post in a series I’m planning on writing on A/B testing. In this post, I’ll lay out a top-level overview of what A/B testing is and why companies do it. In…

Deploying MLFlow on Heroku (with Heroku Postgres, S3, and nginx for basic auth)

mlops

Disclaimer: I followed this guide to setting up…

Notes on Hiring Data Analysts + Scientists

data science

hiring

In the past four months, I’ve been involved in hiring for two new roles at CollegeVine: a second data scientist, and our first data analyst. I’ve learned a lot along the way: Things that work, things that don’t, and things to ask in order to maximize the…

Working With Your Fitbit Data in R

R

data science

fitbitr 0.1.0 is now available on CRAN! You can install it with

Highlights From rstudio::global

R

data science

rstudio::global, this year’s iteration of the annual RStudio conference, was a few weeks ago. Here were some highlights:

What’s New in slackr 2.1.0

R

slackr 2.1.0+ is live! There are a whole bunch of exciting changes that we (mostly Andrie de Vries and I) have made to improve the package a bunch.

A Gentle Introduction to Markov Chains and MCMC

data science

Every other Friday at work we have a meeting called All Hands. During the first half of All Hands a member of the team gives a presentation, which is split up into two…

What I’ve Been Drinking in Quarantine (Plus a Cocktail Lesson)

food + bev

The past ten-or-so months of limited activity due to Covid-19 have…

Our 2021 Big Data Bowl Submission

data science

sports analytics

This post is my team’s 2021 NFL Big Data Bowl submission. My team was made up of me, Hugh McCreery (Baltimore Orioles), John Edwards (Seattle Mariners), and Owen McGrattan…