Cassie Kozyrkov (@quaesita)

Cassie Kozyrkov @quaesita

Chief Decision Scientist, Google. ❤️ Stats, AI, data, puns, decisions. Latin: data = what's given; quaesita = what's sought. Newsletter: https://t.co/YUCreIzHRX

Tweets by Cassie Kozyrkov

Let's take a look at neural networks, this time with a more realistic example... https://t.co/5bS0cVPJoP

Shared by Cassie Kozyrkov at 5/12/2022

Why is logistic regression the most common approach to classification at scale? When should you start your project with this method? And why is it such a good choice for ranking at scale? https://t.co/KIe6ijrhNc

Shared by Cassie Kozyrkov at 5/2/2022

What are all those sigmoid functions for and why you see them all over #MachineLearning and #AI? The answer might surprise you! It has to do with something you'd learn on day 1 of an economics 101 class... This 2min video explains! https://t.co/R7n9wEpRZ0 #DataScience #ML

Shared by Cassie Kozyrkov at 4/28/2022

When should you use linear regression? Some handy advice for identifying situations where linear regression is the best place to start your #MachineLearning modeling journey. https://t.co/wdDwImBAxj

Shared by Cassie Kozyrkov at 4/25/2022

What do all regression models have in common? #DataScience #MachineLearning #Statistics https://t.co/GGOdPplWJu

Shared by Cassie Kozyrkov at 4/24/2022

Statistical errors: * Type I Error = incorrectly rejecting the null hypothesis * Type II Error = incorrectly failing to reject the null hypothesis * Type III Error = correctly rejecting the wrong null hypothesis Decoded for newcomers in the video. ⬇️ https://t.co/0CejrcTscc

Shared by Cassie Kozyrkov at 4/15/2022

Boosted aggregation (bagging) in #MachineLearning is much easier than it sounds: it's all about building multiple models by sampling from the dataset with replacement. The video explains how it works. (Take a look at episode 102 so it makes more sense.) https://t.co/dBCFdHTexg

Shared by Cassie Kozyrkov at 4/12/2022

Explainable AI (XAI) is getting a lot of attention these days and if you’re like most people, you’re drawn to it because of the conversation around AI and trust. If so, bad news: it can’t deliver the protection you’re hoping for. https://t.co/cOnh504xmo

Shared by Cassie Kozyrkov at 4/8/2022

Take a moment to compare your computer-enhanced ability to store, process, and transmit information with that of a typical Ancient Greek. Yeah, you’re basically Athena to them. https://t.co/qRw2xPFxxg

Shared by Cassie Kozyrkov at 4/6/2022

Perceptrons! Such a cool name, but what are they? This video is the first in a series developing the logic for SVMs: * Perceptrons * Maximal Margin Classifiers * Support Vector Classifiers * Support Vector Machines (+the kernel trick with a towel) https://t.co/hsvZfiKQL1

Shared by Cassie Kozyrkov at 3/17/2022

Who's up for some clustering with k-means explained simply? https://t.co/UTm3Yx2Zhx

Shared by Cassie Kozyrkov at 3/9/2022

📊 For your amusement, my irreverent #Statistics glossary. 👇 #DataScience #RStats #MachineLearning #Analytics https://t.co/cyAwB9EpWo

Shared by Cassie Kozyrkov at 2/17/2022

Those of you who've professionally built and launched a #MachineLearning system at scale, I have a question for you: Did you already have an A/B testing platform in place *before* you started building ML solutions? (What's A/B testing? Here's an intro: https://t.co/Hq6hirkXM8)

Shared by Cassie Kozyrkov at 12/4/2021

It's finally here! Your cheeky and fun guide to #AI algorithms: https://t.co/qKkPkNLXyj Happy Thanksgiving, everyone! 🦃 #MachineLearning #MFML #DataScience https://t.co/wN2wFePB9s

Shared by Cassie Kozyrkov at 11/25/2021

Arriving tomorrow! Your fun + intuitive guide to these #AI algorithms: * k-Means * k-NN * Perceptron * Maximal Margin Classifier * SVMs * Decision Trees * Boosting * Random Forests * Ensemble Models * Naive Bayes * Linear Regression * Logistic Regression * Neural Networks

Shared by Cassie Kozyrkov at 11/24/2021

Underfitting versus Overfitting... which one is worse (and why)? https://t.co/gF2JHW8PNM #AI #DataScience #MFML #MachineLearning

Shared by Cassie Kozyrkov at 11/9/2021

3 mistakes that statistically-ignorant data nerds make if they've never had their faith in data crushed by real-world projects: *Thinking you're here to get the right answer. *Working hardest on garbage data. *Forgetting the weakest link in your AI system. https://t.co/FkGHtTyS8z

Shared by Cassie Kozyrkov at 11/5/2021

Question: Why should you care about overfitting in #MachineLearning? Answer: You had ONE job... https://t.co/kPcnDsy1ml

Shared by Cassie Kozyrkov at 11/2/2021

Trick or #DataScience treat? A #Halloween-flavored introduction to analytics! https://t.co/mFoQZA9z5L

Shared by Cassie Kozyrkov at 10/31/2021

How your AI system gets data according to popular belief: it runs around finding data and devouring it all. How your AI system gets data in reality: https://t.co/w2d2ogyd1K

Shared by Cassie Kozyrkov at 10/28/2021

For the intermediates: a type I error and a false positive are different things in the same way a confidence interval and a prediction interval are different things. One has to do with a single prediction and the other has to do with a population parameter/model. 🧵

Shared by Cassie Kozyrkov at 9/29/2021

Good advice for putting the intelligence into artificial intelligence.

Shared by Cassie Kozyrkov at 9/22/2021

3 common misconceptions about #analytics explained: 🙍‍♀️ Analytics is statistics. (No.) 🙍‍♂️ Analytics is data journalism / marketing / storytelling. (No.) 🙍 Analytics is decision-making. (No!) English version: https://t.co/rCAWGnKfJi Spanish version: https://t.co/Ps7cVCuIOo

Shared by Cassie Kozyrkov at 9/7/2021

Having the grace to pivot when new information shows that your previous decision put you on the wrong course.

Shared by Cassie Kozyrkov at 8/24/2021

Setting performance criteria at the beginning - before you even start with data or code! - is an important part of how you keep yourself (and everyone else) safe from horrible machine learning and AI. #MachineLearning #AI #DataScience #Rstats #Statistics https://t.co/8075QR8f6C

Shared by Cassie Kozyrkov at 8/16/2021

It's here!! The most fun you'll ever have in a #MachineLearning course! Making Friends with Machine Learning: Part 1 - https://t.co/R08G8KVfoj Part 2 - https://t.co/kQ4kyj5Qor And here's all the course info: https://t.co/mjdoYf3gP6 RT for your human friends. 💗 #MFML

Shared by Cassie Kozyrkov at 7/15/2021

Common #DataScience oops-and-then-i-learned stories: - Someone was bad at logic. - Someone jumped to conclusions. - Someone hired the wrong team. - Someone had bonkers expectations. - Pointy haired bosses. - Unfriendly tools. - Bad data / GIGO. - Bad metrics. What did I miss?

Shared by Cassie Kozyrkov at 7/7/2021

Machine learning doesn't work without data! https://t.co/Of47EQ7jBY

Shared by Cassie Kozyrkov at 7/6/2021

2 minutes of reinforcement learning, anyone? Making Friends with #MachineLearning continues! #mfml #AI #Statistics #RStats #DataScience https://t.co/lmxAKtsZo1

Shared by Cassie Kozyrkov at 6/20/2021

Think of unsupervised learning as a sort of mathematical version of making “birds of a feather flock together.” #MachineLearning #AI #mfml #DataScience https://t.co/VZUR7T2GVr

Shared by Cassie Kozyrkov at 6/14/2021

If classification is about separating data into classes, prediction/regression is about fitting a shape that gets as close to the data as possible. https://t.co/0rFatsOZth

Shared by Cassie Kozyrkov at 6/1/2021