Data Sciencey Podcasts (Updated)

I’ve been listening to a lot of podcasts this semester since I am driving 1 hour each way to class twice a week, and I thought I’d share some good ones I’ve found. I started out by listening to the entire season of Serial (which I recommend!), then switched to fun and sciencey ones for my commutes after that. I found a few that are data-science-related and wanted to share them here! (the title of each section is a link to the podcast’s homepage)

The Talking Machines

This podcast about machine learning is educational and, though academic, is pretty accessible to people interested in learning more about the field even if you’re new to it. It is executive produced and co-hosted by Katherine Gorman along with co-host Ryan Adams, an Assistant Professor of Computer Science at Harvard.

They start out by interviewing attendees and presenters from the NIPS (Neural Information Processing Systems) conference, including Hannah Wallach and Max Welling, among others.

Today, I listened to this episode with Charles Sutton, who covered some interesting topics such as using machine learning and natural language processing on computer code for tasks such as understanding how different programmers involved in open source projects name variables, and suggesting naming conventions to new project participants. The larger goals of the research were really interesting to me, and Charles Sutton was really clear and easy to understand, even though he was touching on some heavy concepts.

I also like how the show answers questions submitted by listeners each episode.

Partially Derivative

This show has two director/developer/data scientists from Ushahidi, Chris Albon and Jonathan Morgan, who talk about recent data science items in the news, and chat about the implications and add their opinions. I like that they link to the news articles on the podcast site so you can read up on what they’re referring to.

Honestly, I didn’t enjoy this one as much as I did the others. The episode I listened to started out with each of them explaining what beer and wine they were drinking, and how much they had had, and some inside joking and laughing, which already made me cringe a bit (their latest episode is called “morning drinking edition”), but I wanted to hear them out. They talk about plenty of interesting topics, but I had already read most of the articles they referred to via twitter. They had some good insights, including a discussion about Uber and data collection (and data selling) by companies in general, and some interesting food for thought about what all of that means for us in the future, but overall, I found their “bro-y” banter a bit annoying.

However, if you don’t get a chance to keep up with data science in the news, or you enjoy feeling like you’re hanging out with some guys from school drinking and chatting about data science topics, then definitely give it a listen. I imagine a lot of people would like their style more than I did – just not my thing.

TED Radio Hour

TED Radio Hour is an NPR production where they take TED talks and group them by topic, then Guy Raz interviews the speakers and basically refactors the talks so they tell an overarching story as a group and sound good on radio. The episode I wanted to point you to is “Solve for X” because it made me think about math in a fun way, and they do incorporate some talks about machine learning algorithms into this one as well. This podcast is one of my go-tos when I want to learn something interesting that is presented in a fun and curious way.

Invisibilia

Invisibilia isn’t a tech podcast, but does sometimes talk about technology. The episode, “Our Computers, Ourselves” tells the stories of some interesting people that “let technology go to their heads”, I guess you can say. The only thing I don’t like about the show is that sometimes think the intelligent co-hosts Alix and Lulu purposely make themselves come across ditzier than necessary, but overall I highly recommend checking this one out. Also take a listen to the episode “How to Become Batman” that might change your mind about how expectations impact outcomes, even in scientific research.

Snap Judgment

Last but not least, I happened upon NPR’s Snap Judgment podcast with Glynn Washington because it had an episode called “Artificial Intelligence” that came up in a search. It turned out not to be about AI in the sense that developers think about AI, but it has some great storytelling involving human-computer interactions, and really made me think about the human side of all of this technology we are creating. This is a fun one that also gets deep, check it out!

Here are some that I have not yet listened to:

Data Stories (Data Visualization)

Linear Digressions

The Data Skeptic

Tell me if you listen to any of these and what you think, or if you have any additional recommendations – I still have a few weeks of commuting to school remaining, and would love to learn what you love to listen to!

Update 4/24/15:

I have now listened to an episode of each of the podcasts I linked at the end of the post above, and here’s what I thought:

Data Stories: I listened to the episode where hosts Enrico Bertini (who commented on this post!) and Moritz Stefaner interviewed Jen Christiansen from Scientific American. She shared information about her history designing information visualizations for different publications and leading design teams, and detailed the process of creating visuals for Scientific American, which means balancing between making the graphics accessible to a general audience, while also satisfying the scientific readers. Really interesting! I will be listening to more Data Stories in the future.

Linear Digressions: This is the one from Udacity with Katie Malone and Ben Jaffe, and I listened to the 2 episodes about Hidden Markov Models, where they invited a guest (who was a listener!) to explain his work with HMMs. It was really interesting to learn about, with the right balance of technical info and accessibility for beginners, and I will listen to more episodes! The only issue I had with this one was the sound quality. I was surprised it was really hard to hear the guest at times, since they were supposedly in the studio of a company that produces online videos. So, my only suggestion to them would be to work on the volume levels and audio in general to make it more professional sounding.

The Data Skeptic: I listened to the episode about Computer-Based Personality Judgments, where Kyle Polich interviewed Youyou Wu about her research into predicting personality traits using Facebook likes, then he and his wife Linhda responded to the “magic sauce” results based on their own profiles. It was an interesting topic, and I really like how they ask the guest to suggest further reading of their own and others’ work, and then link to everything on their blog. I had a hard time finding info like Kyle’s last name, so I hope they put more info on their website’s home page and make the site a little more inviting, because the podcast itself seems very accessible and fun!

Additionally, I found two more podcasts on Matt Fogel’s post about data science and machine learning podcasts:

Learning Machines 101: I started listening to the episode about “How to Learn Statistical Regularities using MAP and ML Estimation” and honestly I didn’t make it more than 5 minutes into the recording. First, the intro is very long and you don’t get to the content until almost 2 1/2 minutes in, then it sounds like the host is just lecturing by reading from a piece of paper and over-enunciating so it sounds like you are being “talked at”, and it was not engaging to me because of that speaking style. So, I’m sure it has some interesting content, but I just couldn’t focus on it while driving. Sorry!

I haven’t had a chance yet to listen to the O’Reilly Data Show Podcast with Ben Lorica, but the little I heard sounded promising and I’m going to get back to it when I drive to class (for the last week, yay!) next week, so I’ll update here again later.