A Completely Subjective Ranking of Data Science Podcasts

I love podcasts. I would estimate that I listen to between 1 to 6 (yes, 6 😨) per day. Consequently, I have opinions on podcasts.

If you’re still listening to podcasts at regular speed, then I’m not sure I can respect you. (I’m kidding… kind of.) I personally, find that 2x speed is right for me–not so fast that I can’t understand anything, but also not slow enough where I can lose interest due to pacing. Of course, I’m not completely absorbing everything that I listen to, but I’m not too concerned with that–I mostly listen to sports/comedy podcasts, where I don’t need to remember the conversation. In the case that I do want to listen intently to something, then I won’t hesitate to slow it down (probably only to 1.5x, because 1x still feels like glaciers moving).

Aside from sports and comedy podcasts, I do also listen to “smart” podcasts to provide a much needed change of pace for myself, such as those discussing data science. With my self-proclaimed expertise when it comes to the podcast ecosphere, I thought I would make a list of the main data science ones that I enjoy.

This is not a novel idea–there are plenty of other articles ranking/listing data science podcasts. Nonetheless, I provide my list with the disclaimer that it is completely subjective, given from the point of view of a person who does not generally listen to podcasts for entertainment purposes (but can still appreciate a high-quality podcast irregardless of its entertainment value). Also, just for fun, I provide unique “ratings” using model-related error metrics/term for each podcast. These arbitrary ratings are exactly that–arbitrary (although the reader may be tempted to infer some kinds of meaning from them).

In no particular order…

Not So Standard Deviations

  • Personal Rating: 0.72 ROC AUC
  • Average Episode Length: 45 min.
  • Publishing Frequencty: Approx. every 2 weeks

Roger Peng and Hilary Parker have a really nice chemistry on their (shrewdly-named) podcast Not So Standard Deviations. I think that they capture the sense of humor/charisma of the data science community at larger better than any of the other data science podcasts that I have listened to. For example, their “off-the-wall” podcast names, such as “46 - Uncanny Valley of Stickerness” or “Episode 42 - One Piece PJs”, would leave you guessing as to what kind of podcast it is. This is not to say that they’re unintelligible in any way–they blend light-hearted, off-topic conversation with current event news regarding the data science realm seamlessly.

After quickly perusing the web for articles written about data science podcasts, I was surprised to see this one missing from a lot of lists (or not ranked highly). It’s been around since 2015, so I would think exposure would not be a problem.)

Linear Digressions

  • Personal Rating: 0.15 Brier score
  • Average Episode Length: 30 min.
  • Publishing Frequencty: weekly

Katie Malone and Ben Jaffe put out weekly podcasts that have a consistent format–discussion of a single data-related concept, where Katie, a physicist-turned-data-scientist plays the role of a teacher and Ben, a web developer with interest in data scientist, plays the role of the student. This “gimmick” plays well to me as a listener, particularly when I am unfamiliar with the topic which an episode is discussing. While I rarely laugh at the intentional corny joke that Ben (and sometimes Katie) prepare for the open of the podcast, I can appreciate the effort.

Data Skeptic

  • Personal Rating: 0.68 R squared
  • Average Episode Length: 30 min.
  • Publishing Frequencty: Weekly

The Data Skeptic podcast hosted by Kyle Polich provides a mixed bag of weekly content–episodes either feature guests from industry and academia individual interviews, or they provide brief technical discussion of specific concepts on “mini” episodes, where Kyle is joined by co-host Linh Da Tran) on “mini” episodes. While the interview episodes are “hit-or-miss” material for me depending on the guest and my interest in a particular subject, the mini technical episodes are worth a listen regardless of the topic. It follows a teacher-student format (very similar to the Linear Digressions format) in these episodes, where Kyle plays the role of the teacher and Linh plays the role of the student. Linh sense of naivety and her “outsider” perspective echos with me as a listener who may be unfamiliar with the topic that Kyle is presenting.

DataFramed

  • Personal Rating: 0.4 Euclidean norm
  • Average Episode Length: 1 hour
  • Publishing Frequency: Weekly

Host Hugo Bowne-Anderson brings on experts from industry and academia for “deep-dive” interviews in order to explore an underlying question–“What is data science?”. This is the most structured podcast of the ones that I’ve listed, which actually makes it unique in a way. I really like some of the questions that Hugo asks to every guest (e.g. “Where do you see data science going in the future?”, a question that is ironic because it asks the guest to make a prediction regarding a field of study that is nearly all about making predictions) because it helps the listener gauge exactly how different experts view their roles/responsibilities in the data science realm at large. I look forward to seeing how this podcast (which was only started at the beginning of 2018) grows in the future.

The R Cast

Despite the great amount of irregularity in release of episodes of the The R Cast (it has had just over 20 episodes between the beginning of 2018 its inaugural episode in early 2012), I applaud the content that host Eric Nantz (@thercast) provides. The podcast has changed a lot over time, as Eric has seemingly transitioned from concept-centric episodes to guest-focused one. While its content is by no means “superb”, I think that I have reserved a special place in my heart for this podcast simply because Eric is extremely relatable–he just seems like a normal, hard-working guy who loves R and wants to give back to the community whatever he can. While I can’t stand the cheesy video-game music that he uses for transitions, I suffer through it because it somehow symbolizes his “average-Joe-just-trying-to-do-his-best” complex to me.

Others Worth Mentioning

These other podcasts are ones that I have tried out, but don’t really listen to on a consistent basis.

Similar to Not So Standard Deviations in style (i.e. ranges from casual conversation to serious discussion of topics).

Topic-based, which makes it easy to go through and pick episodes at random from its relatively large archive. Probably the most technical show that I’ve listened (as implied by the podcast’s name), which can make it difficult to follow. Nonetheless, high quality content.

Definitely a good show for python enthusiasts. I binged a ton of episodes when I was picking up python. However, ever since I’ve started using R as my main language, I’ve put this one on the back-burner.

Episodes consist mostly of interviews with industry guests. Seems to be more focused on “big data” and enterprise applications. Not super engaging for me simply because it is nearly all business talk.

Focused more so on data visualization than any of the other podcasts. Guest-oriented. Provides a TED-talk kind of vibe.

Share Comments
comments powered by Disqus