I can't believe it's been a year since I released The StatQuest Illustrated Guide to Machine Learning. Thanks to everyone who purchased a copy to support what I do and StatQuest!!! BAM!!! (and if you want a copy, you can get one here: )
It's been almost a year and a half, but I've finally finished writing and editing The StatQuest Illustrated Guide to Machine Learning. It should be available in just a few weeks!!!
#BAM
I just finished writing the chapter on Gradient Descent for my upcoming book, The StatQuest Illustrated Guide to Machine Learning. There's still lots to do, but it feels good to take a step forward.
#BAM
India! The wait is over! The StatQuest Illustrated Guide to Machine Learning is available! Printed in India? BAM! Full color and high quality? Double BAM!! Low price? TRIPLE BAM!!!
Today I finished the chapter on Regularization for The StatQuest Illustrated Guide to Machine Learning (due early 2022). This book is starting to come together!!!
#BAM
Just finished writing the chapter on Fundamental Concepts in Statistics For ML for my upcoming book The StatQuest Illustrated Guide to Machine Learning. Feels good to be done and excited to move on!!!
#BAM
I finished a draft of The StatQuest Illustrated Guide to Machine Learning and am celebrating by recording a StatQuest about DBSCAN, the first StatQuest in 4 months. It's good to be back!
#BAM
!
I'll be the first to admit that my understanding of how to code neural networks is pretty spotty. So today I'm filling in the gaps with
@rasbt
's excellent book, Machine Learning with PyTorch and Scikit-Learn.
#BAM
p.s. Who's picture is on the cover?
Help! I'm writing a book, "The StatQuest Illustrated Guide to Machine Learning", and it, like my videos and study guides, is mostly pictures and more like graphic novel than a normal "words" book. Who (and how) should I ask to publish???
A few weeks ago I lectured at Harvard Medical School on t-SNE and UMAP. I made a special PDF study guide for the students, and now you can have it, for FREE, too!!! BAM!!!
Just finished Chapter 9, Classification and Regression Trees, for The StatQuest Illustrated Guide to Machine Learning (due early 2022). Now on to the chapter on Neural Networks!!!
#BAM
!!!
When I left academics to do StatQuest full time, I thought I'd never again have to deal with The Dreaded Reviewer
#3
. So much for wishful thinking... :(
In Miami for the
@gridai_
offsite, having a great time hanging out with coworker
@rasbt
and his best selling new book, Machine Learning with PyTorch and Scikit-Learn.
#DOULBEBAM
Hooray!!! My StatQuest youtube channel crossed the 900,000 subscriber milestone over the weekend! TRIPLE BAM!!! To celebrate, here are my top 8 videos with over 1 million views! 🧵 1/9
Matrix Algebra doesn't have to hard and cause you to overwork...just learn the essentials for coding your favorite neural network!!!
Brought to you by
@LightningAI
#BAM
!!!
I've been asked to define Neural Network. Here's a draft (in 3 parts):
Neural Network - A long time ago some people tried to create an algorithm that mimicked the biology of the brain and called it a Neural Network. They failed miserably.
A lot of people say you should find a job that you are passionate about. If you can do that, great! Alternatively, you can find something that is important and needs to be done. This may not start out as a passion, but it will become one.
Want to get started programming Neural Networks, but don't want to have to set up a new environment or deal with any hassle whatsoever? Check out this
@LightningAI
Studio and get started with PyTorch right now! BAM!
Norm and 'Squatch are back!!! Today I'm starting my next book, The StatQuest Illustrated Guide to Neural Networks and Deep Learning (with PyTorch + Lightning). BAM!!!
Want to learn how to code Long, Short-Term Memory neural networks in PyTorch but don't want to download anything or install a bunch of packages? Check out this
@LightningAI
Studio that I created and get started right now!
#BAM
!
Another
@LightningAI
Studio: Learn how to code Word Embedding neural networks, then build your own as big as you want with GPUs and deploy, all in the same environment...
As of today I have >950K youtube subscribers and 1M is just around the corner! That means I need a new StatQuest t-shirt design! Help me by posting your ideas below. There will be a vote and the winner gets $100 and a StatQuest shirt!!! BAM!!!
I can't believe it!
@KenJee_DS
and I finally met in person in North Carolina. Great times with one of my long time heroes in data science.
#BAM
!
#66daysofdata
This is the start of a bunch of StatQuests on Neural Networks in PyTorch Lightning. They have given me access to some of the best coders in the world, so the tutorials will be awesome!!!
#BAM
!
Today I am proud to announce that I'm partnering with to bring you StatQuests and tutorials that bridge the gap between doing machine learning on your laptop and doing it in the cloud.
#BAM
@gridai_
My friends
@LearneyAI
made this awesome flowchart of all the StatQuest videos. I love it because you can easily pick the general topic you are interested in and then see all of the relevant videos and their dependencies.
#BAM
If you want The StatQuest Illustrated Guide to Machine Learning, but it's not available in your area...Know that I'm working as hard as I can to solve this problem but it will take at least a month, maybe two. This is a long time, but I appreciate your patience!!!
#BAM
Finally!!! Norm and StatSquatch are on a t-shirt!!! Thanks to
@enderesting
for this awesome design!!! You can get your own in various colors and sizes here:
#DOUBLEBAM
I know this diagram of self-attention for Transformers is a mess, but don't worry, we'll go through it one step at a time soon. I can't wait! (and thanks for your patience!)
If you, or someone you love, speaks Spanish or Portuguese, my latest video (and, hopefully, all future videos) has Spanish and Portuguese overdubs. Just click on the "Gear" icon, then select "Audio Track", and then select the language of your choice. BAM!
Coding neural networks shouldn't be something you're dreading...here's how to use PyTorch and Lightning to code Word Embedding! Brought to you by
@LightningAI
!!!
Day 1 of
#66DaysOfData
Diving into CatBoost, which converts categorical predictors into continuous predictors instead of using one-hot encoding. These are my very raw notes so far...
Day 58 of
#66DaysOfData
Day 11 of UMAP. Here's the main algorithm behind t-SNE, LargeVis and UMAP, with a few notes on key differences. BAM! (1 of 13 panels!!!!)
Technically, a Neural Network is just a Big Fancy Squiggle Fitting Machine. Give it some data, and a Neural Network will fit a squiggle to it. Don’t believe me? Check out the ‘Quest:
For example, you can learn how to code a neural network with this Studio and then just keep building and adding GPUs and then deploy, all in the same environment. TRIPLE BAM
I had a great time last night with the Illinois Data Science Club. The highlight was fine-tuning an LLM and then having it tell me (correctly) why the club is awesome! BAM!
I had a great time on Monday giving a seminar for NC State's Genetics and Genomics Academy
@NCStateGGAGrad
and then hanging out with the students. DOUBLE BAM!
The other day, someone asked me why we use Entropy (or Gini) for decision trees instead of just the raw counts of each class. Good question! And my friend
@rasbt
(Sebastian Raschka) has an awesome answer:
#BAM
Recently, a lot of people have asked me about the most important data science concept for business. I always say "communication", and one the best ways to learn it is from
@GutmanDataHead
's book, "Becoming a Data Head"
URGENT NEWS FOR INDIA!!! If you bought a copy of my book in India and it is black and white, please return it to Amazon. I'm really sorry that this happened. It's my worst nightmare come true. I sincerely apologize for any inconvenience that this problem has caused.
@prmshra
The good news is that I have a video on linear algebra coming out soon that should help cool things down. I also have a few songs about it. Bam.
The StatQuest Illustrated Guide to Machine Learning
by
@joshuastarmer
It takes the
#MachineLearning
algorithms, no matter how complicated, and breaks them down into small, bite-sized pieces that are easy to understand.
Order now
#shroffpublishers
My father taught at IIT-Madras when I was young. Wednesday, Oct 6, I'll give a (virtual) seminar there on the lessons he taught me and how I apply them. BAM!
10am New York, 7:30pm IST
Day 2 of
#66DaysOfData
and day 2 of CatBoost. CatBoost has a unique boosting strategy (called Ordered Boosting) that separates the residuals associated with a row of training data from the trees that were built with that row of training data. These are my very raw notes so far...
If you have time tomorrow, I'll be doing a live stream about Target Encoding (without Leakage) at 11am, New York time (4pm London, 5pm Europe, 8:00pm Pakistan, 8:30pm India)
My little brother: "dude, why are you up so late watching data science videos?"
Me: "no no no, it's
@joshuastarmer
. He says Bam. It's fine."
Brother (five min later): "Oh. He says BAM. That's kinda slick."
#DataScience
#datafam
#ds4a
#data
This coming Tuesday, April 16th, I'll be giving a workshop on how to code and fine-tune an LLM like ChatGPT using
@LightningAI
Studios at
@UNC
with the UNC AI Club. This is free and open to the public! BAM!!!
Day 20 of
#66DaysOfData
Day 3 of Recurrent Neural Networks (RNNs). Here's a simple example of a working RNN along with some main ideas, bam!!! (NOTE: There are 7 parts of this tweet!)
#BAM
Day 3 of
#66DaysOfData
and day 3 of CatBoost. CatBoost does not use normal Decision Trees. Instead it uses Oblivious Decision Trees (ODTs). These are weaker learners (and boosting is all about weak learners) and very fast from a computation side of things.
#BAM
!
On the road to understanding neural networks, I think we're getting warmer...here's the latest 'Quest about a Decoder-Only Transformer!!!
#BAM
#StatQuest
Brought to you by
@LightningAI
!!!
Does anyone know how to report plagiarism to a journal? Figure 3c in this article uses images from my video on RPKM vs TPM
and
but doesn't cite the video.
I just finished recording a song every month for the past 100 months. Thanks to
@aaronkeck
and
@WCHLChapelboro
for helping me celebrate! Here's the Live and Local interview:
Day 11 of
#66DaysOfData
Day 1 of Shapley Values. One step towards explaining machine learning results is calculating Shapley Values. This is the most basic example I could think of (psst! tomorrow we'll build on it).
#BAM
Day 17 of
#66DaysOfData
Day 2 of NLP. I found a great explanation of how Word2vec works by
@JayAlammar
:
His website is full of awesome stuff, so I may just be reading it for the next few days.
Day 5 of
#66daysofdata
and day 1 of LightGBM. If you have a ton of data, building a tree with it all will take a long time. LightGBM reduces the amount of data used to build each tree using Gradient-based One-Side Sampling (GOSS) to speed things up!
#BAM
@kareem_carr
More than calculus, I'd say that trying to teach statistics without every step illustrated leads to a lot of bad statistics classes, where the rules seem really cryptic and the formulas don't make any sense.
I woke up this morning at 4am because I couldn't wait to finish deriving all of math underlying XGBoost. Now I can't wait to make it a StatQuest!!!
#Bam