In this video, we work through a simple example — with real numbers — to demonstrate how to calculate the ROC Curve, an enormously useful metric for quantifying the performance of a binary classification model.
We publish a new video from my "Calculus for Machine Learning" course to YouTube every Wednesday. Playlist is here.
More detail about my broader "ML Foundations" curriculum and all of the associated open-source code is available in GitHub here.
Data Observability — with Dr. Kevin Hu
This week's guest is the fun and wildly intelligent entrepreneur, Kevin Hu, PhD. Inspired by his doctoral research at MIT, he co-founded Metaplane, a Y-Combinator-backed data observability platform.
In a bit more detail, Kevin:
• Is Co-Founder/CEO of Metaplane, a platform that observes the quality of data flows, looks for abnormalities in the data, and reports issues
• Completed a PhD in machine learning and data science from the Massachusetts Institute of Technology
In this episode, Kevin covers:
• What data observability is and how it can help identify data quality issues immediately as well as more quickly resolve the source of the issue
• His PhD research on automating data science systems using ML
• How he identified the problem his start-up Metaplane would solve
• His experience in Y-Combinator accelerating Metaplane
• Pros and cons of an academic career relative to the start-up hustle
• The surprising complexity of the software tools he uses daily as a CEO
• What he looks for in the data engineers that he hires
This episode gets a little technical here and there but I think Kevin and I were pretty careful to define technical concepts when they came up, so today’s episode should largely be appealing to anyone who’s keen to learn a lot from a brilliant entrepreneur, especially if you’d like to found or grow a data science start-up yourself. Enjoy!
The SuperDataScience show's available on all major podcasting platforms, YouTube, and at SuperDataScience.com.
Daily Habit #2: Start the Day with a Glass of Water
This article was originally adapted from a podcast, which you can check out here.
On Five-Minute Friday last week, I introduced the life-transforming concept of daily habit tracking and, if you‘re keen to try it for yourself, I provided an example spreadsheet to get you up and running quickly on your own daily habit tracking journey. If you check out the YouTube version of last week’s episode, you’ll also see that I screenshared the final part of the episode in order to provide a hands-on demonstration of how to configure my habit-tracking template for whatever habits you’d like to track.
Starting with today’s episode, I’ll be recurringly using Five-Minute Fridays to cover the daily habits I track that are most influential in my life. Perhaps they’ll provide food for thought for you or maybe you’ll even try adopting some of them into your own life.
Read MoreThe Confusion Matrix
This video is a quick introduction to the Confusion Matrix, which thankfully really isn’t all that confusing! Understanding what the Confusion Matrix is is key to an Integral Calculus application coming up shortly in this video series.
We publish a new video from my "Calculus for Machine Learning" course to YouTube every Wednesday. Playlist is here.
More detail about my broader "ML Foundations" curriculum and all of the associated open-source code is available in GitHub here.
Interpretable Machine Learning — with Serg Masís
This week's guest is Serg Masís, an absolutely brilliant data scientist who's specialized in modeling crop yields and climate change. He's also a world-leading author and expert on Interpretable Machine Learning.
Serg:
• Is a Climate & Agronomic Data Scientist at Syngenta.
• Wrote "Interpretable Machine Learning with Python", an epic hands-on guide to techniques that enable us to interpret, improve, and remove biases from ML models that might otherwise be opaque black boxes.
• Holds a Data Science Masters from the Illinois Institute of Technology.
In this episode, Serg details:
• What Interpretable Machine Learning is.
• Key interpretable ML approaches we have today / when they're useful.
• Social and financial ramifications of getting model interpretation wrong.
• What agronomy is and how it’s increasingly integral to being able to feed the growing population on our warming planet.
• What it’s like to be a Climate & Agronomic Data Scientist day-to-day and why you might want to consider getting involved in this fascinating, high-impact field.
• His productivity tips for excelling when you have as many big commitments as he does.
Today’s episode does get technical in parts but Serg and I made an effort to explain many technical concepts at a high level where we could, so today’s episode should be equally appealing to both practicing data scientists and anyone who’s keen to understand the importance and impact of interpretable ML or agronomic data science. Enjoy!
The SuperDataScience show's available on all major podcasting platforms, YouTube, and at SuperDataScience.com.
Deep Learning's Neurobiology Origins
Interested in understanding what Deep Learning is? Here's a ten-minute talk I gave last month that visually explains the approach, how it originated in neuroscience research, and how it's now leading the A.I. revolution.
Thanks to Kate Strachnyi and Ravit Jain for organizing the fun "Holiday Book Party" that I gave this talk at. And thanks to Aglae Bassens for the stunning illustrations I used throughout.
Daily Habit #1: Track Your Habits
This article was originally adapted from a podcast, which you can check out here.
In September 2016, Konrad Kopczynski — who happened to be the guest on episode #465 of the SuperDataScience podcast — introduced me to the idea of daily habit tracking.
I appreciate that it’s easy to throw around an expression like “life-changing”, but tracking my habits every day really has been a dramatically life-changing experience. When you wake up every morning and report honestly to yourself on whether you did or didn’t perform a particular good or bad habit yesterday, you open up your eyes to who you really are in a way that our minds otherwise trick us into ignoring or exaggerating.
Read MoreBinary Classification
Last week, I kicked off a series of YouTube videos on Integral Calculus. To provide a real-world Machine Learning application to apply integral calculus to, today's video introduces what Binary Classification problems are.
We publish a new video from my "Calculus for Machine Learning" course to YouTube every Wednesday. Playlist is here.
More detail about my broader "ML Foundations" curriculum and all of the associated open-source code is available in GitHub here.
Data Science Trends for 2022
Happy New Year! To kick it off, this week's episode features the marvelous Sadie St. Lawrence predicting the data science trends for 2022. Topics include AutoML, Deep Fakes, model scalability, NFTs, and data literacy.
In a bit more detail, we discuss:
• How the SDS podcast predictions for 2021 panned out (pretty well!)
• The AutoML tools that are automating parts of data scientists’ jobs.
• The social implications of Deep Fakes, which are becoming so lifelike and easy to create.
• Principles for making A.I. models infinitely scalable in production.
• The impact of the remote-working economy on data science employment.
• Practical uses of blockchain and non-fungible token tech in data science.
• Improving the data literacy of the global workforce across all industries.
Sadie:
• Has taught over 300,000 students data science and machine learning.
• Is the Founder and CEO of WomenInData.org, a community of over 20,000 data professionals across 17 countries.
• Is remarkably well-read on the future of technology across industries.
This episode is relatively high-level. It will be of interest to anyone who’d like to understand the trends that will shape the field of data science and the broader world not only in 2022, but also in the years beyond.
The SuperDataScience show's available on all major podcasting platforms, YouTube, and at SuperDataScience.com.
What I Learned in 2021
This article was originally adapted from a podcast, which you can check out here.
For the final episode of 2017, of 2018, of 2019, and of 2020, Kirill Eremenko — the founding host of the SuperDataScience podcast — provided a long guest episode that he called “1-on-1 with Kirill: What I Learned in the Past Year”.
For today’s episode, to cap off 2021, I’m going to do something similar. Instead of a long 1-on-1 guest episode, I’m doing it as a shorter Five-Minute Friday episode because I had too many exciting guest interviews in the recording pipeline that I couldn’t wait to publish and share with you.
Like Kirill in his annual recaps, I’m going to go over a list of specific lessons. At the end of 2021, I’ve got five:
Consistency Leads to Results
Delegation is the Key to Successful Scaling
Remote Working Works
Real-Life Smiles Are Essential
All Work and No Play Makes me a Dull Boy
Consistency Leads to Results
All right, so let’s start with Consistency Leads to Results. This one is the kind of advice that sounds obvious, yet we often struggle to actually do it. If you want great results at some challenging pursuit, it ain’t gonna come easy. However, if you work at it with unwavering consistency, the gains do eventually come surprisingly easily and the gains build on themselves rapidly.
To illustrate this, I’m going to give you two examples with hard data to back me up. The first is on growing an audience and the second is on weightlifting.
Read MoreIntegral Calculus - The Final Segment of Calculus Videos in my ML Foundations Series
After several months of publishing videos on the Differential branch of Calculus, with today's video we turn our focus toward the *Integral* branch. As ever, applications of this math to Machine Learning remain central.
We publish a new video from my "Calculus for Machine Learning" course to YouTube every Wednesday. Playlist is here.
More detail about my broader "ML Foundations" curriculum and all of the associated open-source code is available in GitHub here.
How to Found, Grow, and Sell a Data Science Start-up
This week's guest is terrifically witty Austin Ogilvie, a prodigiously successful data science entrepreneur. He was founder and CEO of the iconic start-up Yhat and is now founder/CEO of rapidly-scaling Laika.
Austin:
• Was the Founder and CEO of Yhat, a start-up that built tools for data scientists and had a loyal cult following in the data science community.
• In 2018, Yhat was acquired by Alteryx, an analytics automation company.
• More recently he founded Laika, a “compliance-as-a-service” company that dramatically improves your capacity to sell your products.
• Laika last month closed a $35m Series B funding round, bringing the total raised by the firm over two years to a staggering $48m.
In this episode, Austin describes:
• His journey from an arts degree studying foreign languages to teaching himself programming and machine learning, and then bootstrapping a data science start-up into a respected brand and acquisition target.
• His unique take on what makes a great data scientist.
• The hands-on data science tools he finds great value in coding with day-to-day as the founder and CEO of fast-growing tech start-ups.
• His practical tips for growing into a successful technical founder, whether you have a technical background yourself today or not.
Today’s episode will be of great interest to anyone who’s interested in founding, growing, and/or successfully exiting a tech start-up, particularly if you’re thinking of incorporating data or A.I. elements.
The SuperDataScience show's available on all major podcasting platforms, YouTube, and at SuperDataScience.com.
A Holiday Greeting
Five-Minute Friday this week is a holiday greeting from me. Enjoy :D
The SuperDataScience show's available on all major podcasting platforms, YouTube, and at SuperDataScience.com.
Exercise on Higher-Order Partial Derivatives
To cap off an epic four-month sequence of videos on Partial-Derivative Calculus, today's YouTube video features an exercise on Higher-Order Partial Derivatives. Next week, a new topic area begins: Integral Calculus!
We publish a new video from my "Calculus for Machine Learning" course to YouTube every Wednesday. Playlist is here.
More detail about my broader "ML Foundations" curriculum and all of the associated open-source code is available in GitHub here.
Fusion Energy, Cancer Proteomics, and Massive-Scale Machine Vision — with Dr. Brett Tully
This week's guest is Dr. Brett Tully, who leverages his rich cross-domain experience to detail how data science is applied to the fields of nuclear fusion energy, cancer biology, and massive-scale aerial machine vision.
In today’s episode, Brett details for us:
• What nuclear fusion is, how harnessing fusion power commercial could be a pivotal moment in the history of humankind, and how data simulations may play a critical role in realizing it
• How the study of the healthy proteins versus the proteins present in someone with a particular cancer type is accelerating the availability and impact of personalized cancer treatment
• What it means to be a Director of A.I. Output Systems and how this role fits in with other A.I. activities in an organization, such as model research and development
• His favorite software tools for working with geospatial data
• His tricks for the effective management of a team of ML Engineers
• His take on the big A.I. opportunities of the coming years
Brett:
• Is the Director of A.I. Output Systems at Nearmap, a world-leading aerial imagery company that uses massive-scale machine vision to detect and annotate vast images of urban and rural areas with remarkable detail
• As the Head of Simulation at First Light Fusion, he developed state-of-the-art data simulations that could be a key stepping stone toward enabling commercial nuclear fusion reactors
• As the Group Leader of Software Engineering at a major research hospital, he worked to characterize the differences in the proteome — the complete catalog of proteins in your body — between cancer patients and healthy individuals
• As a PhD student at the University of Oxford, he simulated how the cerebrospinal fluid present in our brains flows in order to better understand neurological abnormalities
The SuperDataScience show's available on all major podcasting platforms, YouTube, and at SuperDataScience.com.
Mutable vs Immutable Conditions
This article was originally adapted from a podcast, which you can check out here.
Recently, I had dinner with my wonderful friend Jake Zerrer, who’s a Senior Software Engineer at Flexport, a logistics and supply chain start-up based in San Francisco.
Conversation with Jake is never dull, but I particularly enjoyed a part of the conversation where he brought up an idea for framing problems: He described this framework on the basis of mutable versus immutable conditions.
Read MoreHigher-Order Partial Derivatives
This week's YouTube video introduces higher-order derivatives for multi-variable functions, with a particular focus on the second-order partial derivatives that abound in machine learning.
We publish a new video from my "Calculus for Machine Learning" course to YouTube every Wednesday. Playlist is here.
More detail about my broader "ML Foundations" curriculum and all of the associated open-source code is available in GitHub here.
Data Science at the Command Line
This week's guest is Dr. Jeroen Janssens, a global expert and bestselling author on effectively leveraging the Unix command line as a data scientist.
Jeroen:
Wrote the popular book "Data Science at the Command Line", the second edition of which was published by O'Reilly Media in October
Is the Founder Data Science Workshops B.V., which provides hands-on workshops to global orgs such as Amazon and The New York Times
Is Organizer of the Data Science Netherlands Meetup (3000+ members)
Former Assistant Professor at the Jheronimus Academy of Data Science
Has worked as a data scientist for Elsevier, YPlan, and Outbrain
Holds a PhD in A.I. from Tilburg University
In today’s episode, Jeroen details:
Why being able to do data science at the command-line — for example, in a Bash terminal — is an invaluable skill for a data scientist to have
How mastering the command line is the glue that facilitates “polyglot data science”, the ability to seamlessly borrow functions from any programming language in a single workflow
His PhD research on detecting anomalous events in time-series data
Why LaTeX is a great typesetting language to consider using particularly for creating lengthy documents or technical figures that adapt automatically to new data
How his consulting company, Data Science Workshops, grew organically out of his success as an author
The SuperDataScience show's available on all major podcasting platforms, YouTube, and at SuperDataScience.com.
Ten A.I. Thought Leaders to Follow (on Twitter)
This article was originally adapted from a podcast, which you can check out here.
I was recently asked if I had a list of favorite A.I. thought leaders that I recommended following on Twitter. I didn’t, but that spurred the idea of today’s episode in which I’ll provide you with my ten picks.
My picks aren’t in a particular order overall, but #1 does happen to be my #1 favorite data scientist, and that’s Andrej Karpathy. Andrej is today the Director of A.I. at Tesla, but I’ve been a huge fan since 2016…
Read MoreBackpropagation
This week's video explains the relationship between Partial-Derivative Calculus and the Backpropagation ("Backprop") approach used widely in training Artificial Neural Networks, including Deep Learning networks.
We publish a new video from my "Calculus for Machine Learning" course to YouTube every Wednesday. Playlist is here.
More detail about my broader "ML Foundations" curriculum and all of the associated open-source code is available in GitHub here.