Mama always said: life is like a box of chocolates, and online learning is sometimes like one of those boxes where you don’t know which piece is delicious nougat and which piece is some sort of nasty, coconut-y cream. Well, maybe not your mama. But, it’s a big world, so, surely, there’s a mother out there somewhere who would agree with the sentiment. On this episode, the gang chatted with Google Consumer Insights Analyst Lizzie Allen-Klein about different learning styles and different approaches and options for learning new (and hard!) analytical skills. And there might have been an embarrassing interlude where Tim and Michael exhibited their respective possession of some Y-chromosomes.
Online Learning Resources Referenced
Other References in the Episode
- CXL Live conference
- Google Home
- Snap acquisition of Metamarkets
- Massive Open Online Course (MOOC)
- Cassie Kozyrkov’s posts
- Berkeley’s Data Science blog
- Thomas Bayes’s Grave
- (Podcast Episode) This American Life — The Magic Show
- Cocaine and Rhinestones
01:40 Announcer: Welcome to the Digital Analytics Power Hour. Tim, Michael, Moe and the occasional guest discussing digital analytics issues of the day. Find them on Facebook at facebook.com/analyticshour and their website analyticshour.io. And now the Digital Analytics Power Hour.
02:02 MH: Hi everyone. Welcome to the Digital Analytics Power Hour. This is episode 96. Have you spent all day up to your eyeballs in SQL code? Are you afraid that after all your efforts, you need more education or proof that your education is enough? That’s a quandary that we’re all in, in an industry that’s always changing. And so, all of us must change too. Well, yesterday’s log file analysis skills aren’t gonna get you that sweet data scientist job. And so that’s why we’re here talking about learning as an analyst on this episode. But let me introduce you to my co-hosts. First off, it’s Moe Kiss, who just learned something new today.
02:49 Moe Kiss: Hey, how’s it going?
02:50 MH: Hey, Moe. And next, Tim Wilson, who has actually incorporated the entire corpus of analytics knowledge. Way to go, Tim. You’ve arrived.
03:00 Tim Wilson: Hello all, and fuck you, Michael.
03:04 MH: And I’m Michael Helbling with a strong set of log file analysis skills. Alright. But we needed a guest to round out our discussion and break the awkward pauses that have become the norm for the show. Our guest is Lizzie Allen-Klein. She is a Consumer Insights Analyst at Google. Prior to that, she held product management and analytics roles at companies like Netflix and Metamarkets and now she is our guest. Welcome to the show, Lizzie.
03:33 Lizzie Allen-Klein: Thank you, happy to be here.
03:35 MH: Awesome. Alright. Moe, you and Lizzie met this year and I’d love to hear how you guys got into this topic and then what spawned this conversation. And that’ll get us rolling. What happened?
03:50 MK: Yeah, so Lizzie and I met over in Texas, at CXL, and we got chatting about learning and education. And I think it was a little bit like I was wondering, “Should I go back and study? Should I be doing more courses?” Which I think, as analysts, we’re always all thinking about. And Lizzie talked to me through her journey. I’m really keen to hear you kind of explained a little bit the rationale about the route that you’ve chosen to go down.
04:17 LA: Yeah, I basically just took the windiest path possible.
04:22 LA: And so I ended up in the right-ish place, I would say. I, on paper, I think, nowadays have a background college degree that you might seem lends itself well to going into data analytics. At the time, though, I had no idea. I had no idea what data science was. We didn’t really use any software or anything. We didn’t even… We weren’t even allowed to use calculators in many of my advanced mathematics classes. I had minimal training in terms of SQL, R, Python. I had no training coming out of college, and I just stumbled into analytics. And really what drove me to start learning SQL, which is where I started off, is really the frustration with using tools like Google Analytics.
05:20 LA: And I’m just this sort of person who really wants to know how the sausage is made, except for sausage. I don’t wanna know how that’s made. But for, like metaphorically speaking, I wanna know how the sausage is made. I wanna understand what actually is a user, when I’m looking at users. I wanna understand what a session is when I’m looking at sessions. I wanna understand the exact mathematics that goes into the calculation of any metric. And reading up, myself, on all those metrics and dimensions and everything in Google Analytics, I just felt like, conceptually, there should be a way that I could do more with this tool and I couldn’t. And so I wanted to both fix analytics tools and I also. Wanted to learn more about how data is brought from the Cloud into a user interface for me to interact with ’cause I didn’t know anything about really the cloud [chuckle] at the time.
06:11 LA: I entered a start-up called Metamarkets. They still kind of exist. They were acquired by Snapchat recently. And that is where I really got my first introduction to SQL and it was through talking with my engineering peers. I was actually a product manager ’cause again, I wanted to solve problems with visual analytics tools. I was bought into the dream of the democratization of data for all [chuckle] which now I don’t really…
06:40 MH: Believe in?
06:40 LA: Believe in at all.
06:43 MH: I’m so excited!
06:45 MK: Yeah.
06:45 MH: That’s a… Wait, I think, what we’re forming here, Tim, is this point of disillusionment with data democratization. Like the Gartner Hype Cycle, is like you believe in it, then you get into it for a while and then you’re like… Oh, no.
07:00 TW: I think we just found one of the qualifying questions for our guests though, that’s whenever a vendor reaches out and thinks that their CTO should be on, we should say,” How do you feel about data democratization? Do you have anything to say about it?” And if they gush about it, we’re like, “You’re done, no”.
07:13 MH: Yeah, okay, sorry, back to…
07:15 LA: Yeah. No, it’s okay.
07:16 MH: Wanted to take a little victory lap on that one.
07:21 LA: Yeah, I basically wanted to understand everything about how to query data.
07:30 TW: That jump, ’cause I feel like a lot of my progression has also been… There’s some other team that’s doing something and I’m like, I sort of know what they do, but I wanna play around with it so… But to get from no SQL to dabbling with SQL, was that, you hiding off on the evenings and weekends and trying to figure it out? Or were you talking… You’re like,” I’m the Product Manager. I’m looking over your shoulder. What do I need to install and connect to, to just tinker around with this so I can work with you better?” Or was it neither, those are two extremes and it was neither one?
08:05 LA: It took like two years to get there. Basically, I didn’t… Like I said, I didn’t really know anything about SQL. You go and you Google and I don’t know, for me, I just I don’t retain information and I am very bad at understanding, I guess, mapping a concept to reality unless I actually do it and get my hands on it and see for myself what I can do. Like I said, I studied actual actuarial mathematics. I came out of college having taken a bunch of statistics courses, but I had like the freaking crypt keeper who crawled out of his, he was like 95 or something.
08:40 LA: Teaching me theoretical statistics and he knew nothing about R or anything, and so I came out of that with no actual practical knowledge of how to apply these concepts, and not really understanding what they mean. And so, it was really when I started actually understanding how to query for data that started my fascination with going beyond and then applying some of the concepts I learned way back in the day in college and applying it to the actual data that I was grabbing and manipulating. But to get to there, started off of talking to some of my peers at a start-up who were actually the ones who knew how to query code. They were teaching me the concepts and I had to rely on that knowledge in order to build features and products for this start-up. I had to understand conceptually, what it meant to query and transform an aggregate data and then load it, which is a really hard thing.
09:35 LA: Especially if you miss that aggregation part.
09:39 LA: And I had to understand these concepts in order to make my features and my products usable and also just functional. So you’re not waiting for 15 minutes for a million rows of data to load in what is essentially like a pivot table. There are trade-offs you have to make and you have to understand why you have to make those trade-offs and so you have to understand the concepts of basically ETL. That being said, I wasn’t on my weekends like going and subscribing to different online courses. I would be lying if I said I didn’t check out Khan Academy and Coursera here and there, and I basically tried a few things here and there, but nothing stuck.
10:21 TW: Well, that was my first maybe only, my only multi-week dive into online learning was probably 2012 or so, I probably may… I can probably… Check my LinkedIn profile and find out, was the Coursera course on R, ’cause I was hearing stuff about it, but I had that same, I made it through it. I was every weekend thinking, “Oh crap, I’m not gonna be able to figure this out.” Then I’d get through it, and it would be fine. I wound up passing the class, and then had no… I was trying to learn in the abstract. I actually didn’t have something to do with it and so then like another couple of years passed before I kinda came back to it and I don’t know how much, I like to think that something in that course kind of stuck, but as best I can tell, it wasn’t until I got back and said, “No, I’m gonna do something with it.” I know enough about R, of what I could do with it, but I almost felt like I started from scratch, which, and I struggled the same way when it comes to statistics and machine learning with the digital analytics data making that link. When I’m reading about something, I feel like it’s taken me months at times to actually figure out. I understand conceptually what a random forest is and now what the hell do I do with it with this data that I’m working with? So…
11:46 LA: It’s amazing how when you actually have to apply some of these concepts, your confidence just shatters.
11:53 TW: I got it and then, oh crap. I have no idea…
11:55 LA: I don’t know anything.
11:56 TW: The more I think about it. Yes.
11:57 LA: I literally [11:58] ____ just wasted.
11:58 TW: Oh I have gone through that so many times.
12:01 MK: I do wonder if some of it is the design of the courses, because, I mean, I did the R course through Coursera too, and I really, to be perfectly blunt, I struggled through bits of it a lot. And it wasn’t till after I finished that other people were like, “Oh, actually, that probably isn’t the best course to do.” They teach you a whole bunch of shit about R that you just don’t need to know practically for your job and you will get stuck in that component of the course, whereas, like you probably only need to know 20% of the course to actually be able to do your job really well. And then the other 80%, you would learn as you go and as you need those, need to know those concepts. And I was like, “Oh, yeah, that makes more sense.” When people ask me now, that’s the advice that I give them. But I have found lately after doing some of the DataCamp courses and I found some of them are a little bit more practical because they’re specifically like R for data mining, R for stats, R… So you’re not learning everything you need to know about R. You’re learning it for a particular thing that you wanna do, if that makes sense.
13:07 LA: I’ve been a big fan of DataCamp.
13:08 MK: You are or you’re not?
13:10 LA: I am. ‘Cause I like being able to learn, I guess, techniques that I can apply. I think that stepping away from formal education, if you want to understand how best to learn online, or other ways offline, know thyself. Look back at how you best learn over time, like how, like in your experience, what points in life work did you do really well when you were studying, what points did you not do well. I can tell you, I can go all the way back into freaking elementary school and tell you what I did and did not do well because I thought about it so much, ’cause I’m like, I try and figure out why is it that try out Khan Academy, try out Coursera. I google what, how best to teach yourself data science, and you see all these Quora posts about people who list a thousand different links to a thousand different courses, and you’re like, “Okay, bye life.”
14:08 LA: And no, it just, it didn’t work for me. I’m like, why doesn’t this work and why am I not that person who goes out and dedicates their entire life to perusing these thousands of different links to different courses? And for me, it was just because I need both practical application. I need to feel the pain because it’s like my life, not my life, my work life depends on me getting this concept right. I need to be able to use a concept on my own data. That, and I actually do need feedback and consultation from somebody. I can’t just submit my work to the Cloud and then just keep on trying until I get it right. At some point, I get stuck. And you either google the right answer and figure it out, or you can… Like if it’s…
14:57 LA: Certain programs have resources that you reach out to, or forums that you can talk to different people and learn from them. And that’s something I’ve learned back. I used to like frigging, like almost flunk out of math, [chuckle] and I’m a math major. But I struggled because I needed help and I’m not too proud to admit it, and I’m not too stubborn and I don’t have that much of an ego to be like, “Oh, I can figure this out on my own.” I’m just gonna go and ask the experts and learn from them. And I apply that everywhere.
15:31 MH: I was gonna highlight what you said because I think it’s actually important for people to hear, which is, your learning style really factors into this a lot. And so, if you have an experiential style or kinesthetic style, like getting your hands on it and doing it is like the only way. Other people can read a book about it, and suddenly, sort of pieces of it makes sense to them because they have a visual or auditory style depending on, you know? Obviously, the people who listen to our podcast are auditory learners, maybe. I don’t know. Or just big fans of the show, whatever, not learning a darn thing.
16:07 TW: They just…
16:07 LA: One or the other.
16:09 TW: But they keep insisting seeing it on Reddit and they just.
16:11 TW: Yeah.
16:12 MH: It keeps their…
16:12 TW: Yeah, yeah, yeah.
16:13 MK: Yeah.
16:14 MH: Our huge new social strategy is all Reddit-based. No, but I think that’s a really good point that you made about identifying your style and as an analyst, you dug deep into your own experiences and you looked at, well, where did I excel educationally, where did I struggle, what were the commonalities, and you picked it apart. I think that’s really good tips.
16:40 TW: How did you go about… I’m sure in some cases, it’s obvious, there’s this one thing and there’s this one course, but I mean take Python for data science. Shit, there are probably 15 online courses between self-paced tutorial on Code Academy, all the way through entire programs on edX. How, when you’ve gone to take something, how have you researched or figured out this is kind of the right thing, or do you just start with one and then bail if it’s not? How did you go about filtering down to what’s the one thing you’re gonna do now that’s not all-consuming?
17:24 LA: I learned SQL kind of on the job.
17:28 TW: Yep.
17:29 LA: But… And that job being a product manager on Netflix, because we were launching new Numetrics and I needed to QA them.
17:39 TW: I learned the LOTUS script on the job that way. And I think you got a much more sustainable broadly applicable skill, but… So, but putting that aside, that was kind of the like, oh, inching along with an immediate need, but it’d be…
17:57 LA: Right, yeah.
17:57 TW: So…
17:58 LA: But learning SQL was literally just pissing off my DBAs at Netflix for a long time. They’re going to put a limit clause, my God. [laughter] It wasn’t until I basically… When was it? It was the break between Wizeline and joining Google on contract, that I dedicated myself to fine-tuning my skills and learning. I already knew SQL, but going, pushing beyond that, and learning R and Python and etcetera, and I knew I’d struggled. I wanted Code Academy to work. And I should say as a blanket statement, my criticisms of any online program isn’t that it’s a bad program. It just didn’t work for me.
18:47 MK: Yeah.
18:48 LA: And Code Academy seems like “Oh, you’re using what seems like theoretical over practical data,” and you’re chugging through it, and you’re just constantly submitting assignments, and getting graded on it. You pass-fail, essentially. But pass-fail doesn’t work for me. I need feedback and I need to understand how to do something better, and I’m also very goal-oriented. I need to be working towards something.
19:12 TW: Other than oh, this question, this question, fill out this, click the right thing when you’re… Yeah.
19:16 LA: Yeah. I need to know where I’m going and why. And so, basically, in that tiny little break between my jobs, I decided, I was gonna start Udacity. And I did trial every… Almost all of these do, like if they’re a paid program, they do trials. I take advantage of the trials to see if it sticks for me and for me, Udacity really did stick for me. There’s definitely some downsides to it, but what I liked is that I felt like I knew what program I was working on. I felt like I knew I was gonna achieve when I was done with each module, and I felt like my homework is getting graded. [laughter]
19:57 LA: Not just pass-fail, but actually, I was getting feedback. The first time I got a feedback, I submitted something and I knew it worked technically, but they came back and they were like… There’s a way more efficient way to do this and I was like, “Yes, that’s what I need to know.”
20:10 MK: Very nice.
20:12 LA: And ’cause I can find 50 different ways to solve a problem, and 49 of them will be massively inefficient and I’ll be proud of the fact that I found massively inefficient ways to solve problems, just because it’s creative, right? [laughter] But I also need somebody to tell me, “Hey, there’s a way, there’s a shorter path to success.” And so, I think that sort of feedback is really helpful, and not the sort of thing that you can get just Googling the answers to problems. It’s trying to figure out what’s the logic behind it, I guess.
20:46 TW: Well, I definitely have seen people who have taken courses in R and their code is abysmal and they clearly have not gotten feedback from somebody who knows how to write code. Oh that you…
21:01 LA: I’m sure that I… [laughter] Yeah, it was funny because I didn’t even… I didn’t know tidyverse. I was like, “Oh what’s this tidyverse thing about?” I remember looking into it and be like, “Oh, I actually learned that as part of Udacity.” For me, that was the… It’s a learning… It’s particular learning style that I prefer and that’s what worked for me. Now, I should say that in parallel, I was also starting a new job at Google, and I had the freedom to actually try and apply some of these concepts to what I was working on. I was managing experimentation for Google Cloud Platforms website, and it was all manual pulls out of GA input into a spreadsheet and use particular calculations for your statistical tests. And I just was like… Aah… Uhh, nah. And so, I basically decided to start trying to pull the data through SQL and figuring out how to actually analyze the results of experiments using R, that I was learning in parallel through Udacity.
22:13 LA: And I feel like that is what actually made it stick for me because in my mind, I was helping myself do better in my job and I’ve realized that is not necessarily a luxury that everybody has. But if you are in a position, I mean the Klein in my last name is working mostly out of Excel as a business analyst for Yay and I tell him all the time, “If you see a window where you could find the space to try and do things more efficiently, and in parallel, work on learning SQL, Python, R, do it.”
22:48 TW: But that’s… For one, I think there does have to be some intrinsic motivation for the analyst doing it.
22:54 LA: Yeah.
22:54 TW: ‘Cause I feel like I found that by saying I’m gonna basically split my time with… I know I could do this better. Would commit to it and say I’m gonna have to be dedicated. I get a little irritated with people who are like, “Well, I don’t know that. How within my 8-hour workday are you gonna let me do all this?” And I’m like, “Well, this is not the job or the company.” They should enable you to be successful, but if you literally have to fit it perfectly into your day-to-day task, while I’m doing this repetitive thing with Excel, and oh my God, doing the statistical calculations is kind of jumping through a bunch of contortions. And look, I found on Stack Overflow, that’s literally one function in R… Oh, crap. Now, I have to pour in all the effort to figure out all the other stuff to get to use that one little function.
23:49 TW: But I think to, complementing Michael, when you said different learning styles, I also feel like being honest with yourself, like, do you wanna be uncomfortable? Are you ready to beat your head against the wall? ‘Cause…
24:00 LA: Yeah.
24:01 TW: SQL, Python, statistics, machine learning. None of those at this point have a… You can. You can go through Code Academy and feel really like, “Oh, I think I’ve learned Python ’cause I went through that and that was pretty easy.” And then, you’re like, “I import what?” “Import pandas as PD.” “Huh?” So I think that… I mean, I guess my experience has been that I’ve gotten way more comfortable with being pretty frustrated and beating my head against the wall. And it’s more rewarding when you get through it and you beat your head against the wall 27 times, and all of a sudden, the wall starts to, you’re doing it… Instead of five times a day, you’re doing it twice a day.
24:43 MK: I don’t even think it’s that you need to beat your head against the wall to get progress. I think it’s more that you need it to remember. For me, I found when I started learning R and SQL at first, I was pretty much on my own and I didn’t really have anyone to help me, and I’d get really frustrated. And then more recently, I’ve had a really good network of people that I can ask and I got lazy and I would just ask them. And one of them gave me some shit over drinks and was like, “Hey, Moe, you know you’ve asked me that three times now.” And I was like, “Whoa. I’m doing too much asking. I need to do more beating my head against the wall.” Now I have a new thing where I don’t ask anyone. I give it an hour and if I’m still stuck after an hour, then I’ll ask. But I try and self-solve because otherwise, I don’t remember. If someone just comes over and changes the line of code, I don’t remember how to solve that issue the next time versus if I spend the hour beating my head against a wall and reading like five blogs. And even if I still go ask someone, I remember how frustrated I was about the issue and then I remember the solution to the issue.
25:50 TW: See, I remember all the dead ends and I don’t remember which one actually took me to the solution.
25:54 TW: So then, I’d just repeat the same pointless exercise. I’m like, “I’m back on the same fucking post again and it did not help any of the last three times.”
26:04 LA: I think that’s part of the journey. I don’t know. For me, it’s not all intuitive to me. And so, I need to figure out how to make logical connections between these different minute problems that I’m solving. I need to figure out how to take this really abstract complex problem and break it down into sub-parts, and then translate those sub-parts into code. And if you can’t do that, then you can’t… It’s only when you figured out that connection that you can repeat it. Again, it’s a knowing thyself sort of thing. For me, I always… I was one of those geeks that would take the homework and repeat it over and over and over again like a fucking crazy person.
26:52 MK: Wow. [laughter]
26:53 LA: Because I figured out early in test-taking, in public education, that literally most of the time, teachers just take some subset of problems that they’ve offered in homework and swap in the numbers. But if you know the structure, you can figure out. It’s more about learning how to solve the problem than the actual numbers itself. And so, for me, it just trained me to learn how to solve the problem. And I think when it comes to taking abstract, like business problems that you need to solve in real life, breaking that down and figuring out how to solve it through code is just another word problem. Literally, most of today, I was just looking at a word problem and trying to figure out how to translate into SQL and driving myself crazy. And I still haven’t figured it out. But that’s part of the struggle. Like you have to beat yourself up for a little while and either you have your epiphany or you ask for help. If you don’t do one of those two things, you gotta time box yourself. You can’t just let yourself spin out of control and burn out.
27:58 TW: But what happened… With the Udacity, or when it comes to back to the online learning… And Udacity is not one I’ve experimented or done any courses with. What’s the mechanism if you’re beating your head against the wall with a problem set or an assignment, what’s the… Like some platforms, seems like they’ve got like, “Oh, there’s forums.” If it’s classes that are scheduled, some really don’t have a… Like you’re just stuck until I guess you google for the answer. But what about that part of within a course getting help?
28:32 LA: With Udacity, they have forums, but I will be honest. I actually did to the paid program ’cause they have Nanodegree programs and I did a Nanodegree. I’ve done technically two. They restruct… It’s a long story. But I definitely did Intro. I did the Intro to Programming. It was a nice intro to like Python and yada yada. I did the data analyst Nanodegree, but I didn’t actually finish it, and I will explain why later, but they do have forums. I think, with the paid service, you can also reach out to a teacher or whatever. I don’t know what you call them, online teachers.
29:10 MH: A docent.
29:10 LA: A person who knows more than you, whatever.
29:12 LA: Sensei, whatever. But I don’t know if the forums are accessible regardless of whether you’re in a paid program or not, but that being said, I am one of those who googles everything. And I honestly think 80% of the battle with googling is figuring out what to google.
29:32 MK: I agree. I agree.
29:34 TW: Yes.
29:34 MH: Yes. Absolutely.
29:36 LA: Like it’s literally, if you could take one thing away from online courses, it doesn’t matter which one. It’s knowing the terminology and then being able to apply that to your googling.
29:48 MH: Yes, I feel like it’s one of my biggest qualifications actually is the…
29:53 MH: It goes in two parts. First, googling effectively and asking questions the right way. Those two things are our super powers.
30:02 LA: Exactly. You can’t ask your business question, right? Either, there’s too many things in that question… Terms in that question that nobody gives a shit about and no one’s gonna… It’s just not gonna show up, the answer is not gonna show up. You have to know the technique terminology, that’s what you need to know. And once I learned that and I do think that online courses did help me at least learn that, it gave me a path to debug like 80% of my code.
30:36 MK: See, my biggest issue though is whether or not a course has an offline capability. That to me… I’ve actually really been struggling with that because I love DataCamp. I think they structure things really well, but if you wanna watch it offline, you have to go through and download every two-minute video. And so, for like a chapter, there is 12 videos you have to do it and I’m like, “This is a shit show.” I’ve actually swapped to Coursera because you can download a whole week’s worth of work, do it offline on a plane, and then come back and to me, that’s like…
31:10 TW: Is it really about a plane, Moe? ‘Cause I feel like we need internet in order to record this podcast and we’ve kind of experienced some of your challenges on the internet.
31:18 MK: What’s this offline you speak of?
31:20 TW: You’re partially offline…
31:22 MH: Just by dint of being in Australia. Is that what you’re saying, Tim?
31:26 TW: I think it’s her house but…
31:28 MH: Okay.
31:28 MK: It’s fine. My house is fine.
31:31 LA: I don’t live in the offline.
31:33 TW: But I guess we didn’t… Like with the Udacity, in Intro to Programing, is it multiple courses? Is that like a… All of the MOOCS tend to have a, “Oh, here’s a program, you take these series of courses.” Was Udacity like that? How much of a plunge I guess, were you taking with the Nanodegree? Was it, “I’m gonna take this and I can and most likely quit tomorrow?”
31:56 LA: At the time, they had an incentive where you get half of your money back if you finished within 12 months and so for me, I was like… I don’t know if they still have that actually. I don’t think they do, but at the time, that’s what they had. And so that was, for me, I was like, “Game on. Let’s do this.” [chuckle] And so, spoiler alert, I didn’t.
32:16 TW: Well, then, how long did it take? What kind of a commitment?
32:22 LA: It would have taken me under 12 months, but I stopped and I didn’t get that money back because I actually… In my job, I actually… Things just took off and I just I guess, I hit that moment where these concepts stuck and I realized I was applying them in my day-to-day and not only that, but I was helping others and I think I felt like I could now… I basically, I felt like I could… I kinda stopped right at the ML part and I was like, “I don’t need to go there right now. I need to go there eventually, but I don’t need to go there right now. Right now, I need to focus on my work, and now I’m applying a lot of these concepts in my day-to-day work.” And I just felt like I could learn more on the job than I could off in an online course at that point. That was just a personal decision for me.
33:14 LA: That being said, a year passed and I came to the… One of many crisis points where I was like, “Okay, I don’t know ML. I don’t.” [chuckle] Do I need to actually… And at the time, I wasn’t like, I don’t wanna sign up for another online course and/or pay a bunch of money to an online course that, where I just learned packages and I don’t understand when I should apply them, I guess. I had learned enough at that point that I realized that just learning packages and being like, “You can learn ML in X weeks.” I just felt like that was snake oil a little bit.
33:56 LA: For me personally, I feel like you can, but there’s all this other stuff that you’d have to learn as well on top of it and for me, I wasn’t convinced that signing up for another paid online course was the right course for me. I decided I actually needed to challenge myself professionally because again, for myself, I knew that the best way for me to advance my techniques is to put myself in extremely uncomfortable situations in work and turn the heat up and rise to the challenge or fail and learn.
34:31 MH: Yeah. I wanna kinda turn the conversation a little bit to a couple of other ways of looking at this. We’ve talked a lot about our sort of personal approach to learning. To what extent do the companies you’re part of, and this is for any of us, play a role? ‘Cause we kind of said… I think we sort of implicitly said, “Hey, you’re gonna have to spend a lot of your own time doing this if you wanna get better.” But where does the company come from and what is their role in this in enabling people? ‘Cause I think, obviously, everybody who listens to this show is an awesome self-starter who’s on their way to great things.
35:15 MH: But…
35:16 TW: Or at least they’re on their way to work.
35:18 MH: Yeah, which hopefully, well… Great things. But where does the… Who you work for, what could they be doing or what have you seen, they’re like, “Wow, you know what, that was key in that role.” or “Man, I gotta get out of this job because it’s holding me back from doing the things I know I need to do to learn and advance my career.”
35:41 LA: I think if I could go back in time, to my younger years in my career, I think I would have kicked myself in the ass to not be so insecure about asking other people to help me learn new skills. I don’t think I’ve ever been able to take time out of my day, my actual work hours, to like, look at an online course. I don’t think I’ve ever had a company to actually like explicitly say, “You get X hours per week to just study.” I think I would go back in time and have made connections with people to work with them, perhaps outside of work to help learn new techniques. I think it’s somewhat unrealistic even at the best companies to expect that you’re gonna be able to take time out of your eight-hour day, if that’s your situation, to study up on new techniques. That being said, some companies allow that and encourage that and you can make that into… You can reason it out if you’re trying to solve a problem that the business will benefit from.
36:44 TW: I’ve never worked at a company or seen a company that tries to push their people to learn where it is not a colossal waste of money on the company’s part. [chuckle] Because I think, I strongly believe that it’s great for companies to not get in the way like in support. “Hey, if you want, we’ll meet you halfway with tuition reimbursement if you wanna go back to school or we’ll support that.” But I mean, I’ve got some infuriating examples of like a basically a Design of Experiments class from Ohio State and I was working at a company in Columbus and they said, “We’re gonna have the professor come. You’ll get time at work. You just go down to the basement. You’re taking a college level class in Statistics.”
37:33 TW: And it was almost like people are like, “Oh, great. I get to do everything in my eight hours.” And they weren’t remotely engaged in the content like ’cause they kinda had no skin in the game. They were trying to look at it as, “Oh, the organization is doing this.” And even companies that have multi-faceted ways of learning, like trying to push employees to learn if they’re not intrinsically motivated to learn and challenge themselves. And I think at most companies, you’re gonna wind up with spending a lot of money pushing people that’s not really gonna get you or them anywhere.
38:09 MH: Yeah.
38:09 TW: But I don’t have striking views about it or anything.
38:13 MK: I have a different experience in that, my company, we have learning day once a month. We have pet project day once a month. And I think it’s actually one of the benefits about data and analytics team sitting in the technology department, is that we sit alongside engineers and our engineers sure as shit need to keep learning and keep their skills up. And it’s expected that our analysts and data scientists do the same. I, 100% would not have an issue if I was sitting at my desk one day for three hours on Coursera or DataCamp or whatever education provider, no one would have a problem with that at all. Yes, there is an issue.
38:48 TW: That’s literally not at all what I was saying. Like… Did you…
38:51 MK: No, I’m not…
38:52 TW: Did you completely miss the point?
38:54 MK: No. I haven’t finished.
38:56 TW: Okay. [chuckle]
38:57 MK: Haven’t finished. What I’m saying is, I feel that in my like eight-hour day or whatever, it does sometimes become a nine-hour day because I’m choosing to do that learning or a 10-hour day. It’s not like it’s all on work time, but the one thing that I do struggle with is, then there is an expectation that everyone can learn anything like all these online courses, if you don’t have a particular skill, it’s kind of expected that everyone can just pick it up. And some people just don’t have the interest. And some people, they don’t wanna learn that stuff, but there’s an expectation that if there’s any skill you need, there’s just this abundance of online courses and you can teach yourself. And I’m not good at teaching myself. I’m like… I feel like Lizzie and I learn very similarly. I need feedback. I need people to ask for help. I need to have practical examples. I mean, we do have some like lunch and learn sessions, but I still feel like I don’t… I just… I can’t sit on my own for four hours and have everything sink in. Like it’s just…
40:00 MH: No, I think that’s interesting, Moe, the concept of like there being an expectation that because it’s available, that everybody knows it. I know a few years back, I rolled out the Coursera Data Science certificate at our company and offered to reimburse anybody who wanted to do it. Not a single person did it. And I was like, “Oh, I guess people didn’t want to learn about data science.” And then later, after it’s like then people would be like, “Hey, I wanna take this data science course.” And I was like, “You didn’t even do the one that we offered to pay for like what’s… ” I do think there has to be both sides of the equation, right? There’s gotta be opportunity. I think it has to happen. Volitionally, it’ll only happen if the individual makes it happen ’cause to your point, Tim, you can sit in a class and just snooze. But I think if you’re gonna learn new stuff, it’s gonna be because you sat down and did it, which I think is interesting. Okay, so switching gears again.
41:04 MH: There is also a perspective that I think we could lightly tackle ’cause we’re actually getting pretty close on time, so let me just open up a real big can of worms, which is, what about how this applies across genders? Like as men and women, I think there may be different expectations, certainly different experiences. And so, I’d love for us to maybe chat a little bit about that if you’ve noticed or seen any of those, as it pertains to your training knowledge growth experience, all those kinds of things.
41:40 MK: Lizzie, I actually think we might have talked about this, but I have quite a few friends that are engaged in like say, they’re doing the Coursera Masters of… Or Data Science specialty or I have another girlfriend who’s doing a Masters of Data Science. And a lot of them have expressed to me that they really feel that they need that degree and that piece of paper, something that they can put on LinkedIn or on their resume that says “I’m qualified in this” because otherwise, they’re afraid they won’t be taken seriously, or that their skills won’t be recognized. And I can understand their perspective and I really sympathize with it, but it’s really interesting to me because you are such a confident person and you’re so focused on your learning that the point of it is to learn. It’s not about the bit of paper. Yeah, I’m interested to hear your thoughts.
42:33 LA: Yeah, I struggle with the desire to want the degree for similar reasons where it’s like I just wanna be that much more competitive and show to a very male-dominated industry that I am that much more qualified. I can absolutely relate to that, and that’s something that I’ve felt over and over again where I have had these moments where I’m like should I just put the rest my life on hold and go back to school and get a Master’s or go back and get a PhD.
43:06 TW: It seems like that putting your life on, going back into that degree for you would be kind of inefficient from a learning perspective. There’d be a lot of peripheral stuff that just from a raw knowledge would be challenging. That’s the opportunity cost of… I would then have this… I would hope that the industry is starting to evolve and get enough of a clue that it starts to become, yeah, if that company with that group, or that particular asshole, that insecure dude… And it’s just like, “Well.” I would hope there would be more organizations, and even as education is evolving and people are realizing that the online learning comes up, that that may be a barrier if someone is saying where I am right now, wow, if I don’t have a degree, then I don’t know that culturally, it’s gonna be acknowledged that an alternative might be, “Well, maybe I shouldn’t be here. I can spend a year kicking ass with online, tailoring to what makes me effective and be able to show and demonstrate my skills.” And I’m completely cognizant that this is now the dude like chiming in on this particular topic. I mean it is… And it would just be… It makes me cringe.
44:30 S?: Tell us more.
44:30 TW: It makes cringe.
44:33 TW: Good Lord. Nicely done.
44:36 MH: I think what you meant to say, Tim, was…
44:38 TW: You know what? Yeah.
44:40 TW: If you could explain… Oh my… Yeah.
44:45 LA: Let’s get another man in here that makes a point. Let’s double mansplain this.
44:49 TW: Right. But I mean… I mean it’s…
44:50 LA: Just kidding.
44:51 TW: It is so depressing that if we’re not feeling like, not that the end is in sight, there’s at least progress that we’re not having to do literally in absolute terms, not what makes the most sense in order to advance in the way that reasonably someone should be able to advance. I’ll stop now.
45:14 LA: Yeah.
45:15 MH: Yeah. No, all I was gonna comment was, I think I’ve felt that need for validation or sense of do I need this to be competitive. And so, in some small way, I think I understand. But certainly, I don’t think I can see it from the perspective of there being this real layer of sort of expectation of like, “Oh this guy’s got these things. I’d absolutely hire him over her who doesn’t seem to have those things when all other things may be equal or even advantageous for the woman.” I think that’s certainly something to think about.
45:54 MH: Okay, one of the things we do on the show. No, Moe… What?
45:58 MK: You need to let Lizzie finish her comment before we roll into the…
46:05 LA: Literally two men just took over our conversation about gender and how that factors into decisions about progressive education. And…
46:14 TW: Alright, we’re backing up. I’m sorry. I missed the signal very badly. Though…
46:21 LA: Oh, okay. I was comically letting you guys do this until I could make a crack at you guys, which I did.
46:26 TW: No, no, bring it back. Bring it back.
46:28 LA: Let’s bring it back. As I was saying, I have felt… You made a comment about my confidence. We all have our moments, and I have definitely felt that I need a PhD in order to compete with the men. And I always felt that way. And no, I guess I’m just stubborn, like I don’t wanna just give into that expectation. And I think the best advice that I’ve gotten is your portfolio speaks for itself. And so, if anything, work on your portfolio. It’s very, very difficult to compete with a PhD, but if you have practical experience solving business problems, that speaks volumes more than building something for theoretical purposes.
47:18 LA: I know that seems like easier said than done, but for me, even though I may not be able to compete with people in certain ways who have a computer science degree, I know that I bring a value to the table which is that I can solve problems and I can make those connections very fast in terms of how to find a problem, and I know code. Basically, leverage the strengths you already have. Learn how to round out the skill set and you’re probably already working with a lot more than someone who spent most of their life in academia.
48:00 MK: I just… No. I’ve just had an epiphany, an actual epiphany.
48:04 LA: Not that it’s setting your life in academia is bad, by the way.
48:06 MK: No, my epiphany is…
48:09 LA: Sometimes I fantasize about it, but…
48:10 MK: You’re totally right, though, on the, why don’t we do this more? Of having the portfolio of like, here is an example of some MySQL code and here are the comments on the insights that I found and shared with the business. I was answering this business question. Here is some of my R code. Lots of companies do tests, but why don’t we do… This is genius! Genius!
48:32 LA: You just build out your GitHub, man.
48:34 MK: Yeah.
48:34 LA: Who cares what your gender is if your code’s way better than anybody else’s? It’s way more efficient and you actually have actual insight coming out of it. Like who gives a shit what’s between your legs.
48:45 MK: Yeah, nice.
48:46 LA: I don’t know.
48:47 MH: Awesome.
48:47 LA: Lots of people do.
48:48 MH: All right. On that note, now we do have to start to wrap up. This is really great. Tim will edit out all of the parts where we all look like jerks.
49:01 TW: No, I’ll just stay in looking like a jerk.
49:02 MH: Okay, and then let’s jump to last call. One of the things we like to do on the show is just go around the horn and share anything we found recently that is of interest, or we think might be interesting to our audience. Lizzie, you’re our guest. Would you like to go first?
49:20 LA: Sure. My last call is basically a plug for the blog/writings of Cassie Kozyrkov, which, disclaimer, works at Google. I do not know her personally at all, but she is the chief decision scientist at Google. That’s her title, which is a kickass title. She has a way of writing about very technical concepts in very relatable ways. And I started following her on Twitter, have been following her writings for a little bit recently, and just am kind of enamored with the way that she communicates technical concepts. It’s very tempting for people to write about technical concepts and go straight into the weeds. But she keeps it higher level, but it’s still something that can stick with you by using analogies and just actually using humor baked into all of this, like comparing ML and AI to an island of drunk co-workers. It makes sense when you think about it, not like a magic box, and instead, like an island of drunk co-workers that you’re trying to train to identify things. It’s actually like, oh yeah, there are limitations to this and there’s an actual application to… Anyway, so I just think the way she communicates very technical data science concepts is, for me, extremely accessible, which I enjoy.
50:39 MK: You’re gonna make my last call look really shit now, thanks.
50:43 MH: Hey Moe, would you like to go next?
50:45 MK: Oh, I don’t wanna go next.
50:49 MK: I actually got pointed to, I didn’t know Berkeley had an official blog of machine learning, but they do, and one of my colleagues Si Hill pointed me to it. And I actually started trolling through. And I, again, I found it really good ’cause it was really easy to, like they explained concepts really easily and they’re really easy to follow, but I feel like now I’ve got some other reading to do. Maybe read Lizzie’s first.
51:16 LA: No, read ’em all.
51:19 TW: I found that Quora… My last call’s gonna be… I’m gonna start a Quora question and answer it myself where I’m just gonna list the stuff you guys are recommending to read. It will take so many hours.
51:28 MH: Yeah.
51:31 TW: Michael, you wanna do your last call?
51:34 MH: Yeah, actually, my last call is a little bit weird, but it’s the only thing in the last week that I was just really fascinated by. There’s a company named Fairygodboss and they are a startup that works with corporations to help promote a gender equal workplace and then creates job listings for women. And so, they’ve done a ton of research into how women evaluate roles at companies and especially in tech. And they’re doing a lot of work to help companies realize what they can do to make the workplace more accessible and equal. And then helping those companies generate… I mean it generates a lot of positive things for those companies, including being able to recruit highly qualified women with their company. The company’s name is Fairygodboss. It looks really cool. I met somebody from the company last week and then I got a chance to talk to them for a while and then check out their website. Nothing to do with analytics, but if you’re a woman in technology, you should definitely check out their website, and if you’re a guy, you should also check out their website because you’ll learn something. Hey Tim, what’s your last call?
52:44 TW: Mine is gonna be much off the… I’m gonna try to tie it into analytics and it’s gonna be a little bit of a winding road, but I was actually, got to meet Donal Phipps a little while, a few weeks ago. And I had never met him in person but we’ve exchanged a lot of, collaborated quite a bit through the Measure Slack, and GitHub and Twitter and elsewhere. And so one, it was just kind of fun that because it happened to be where we were meeting in London, he was like, “Hey, Thomas Bayes is very near there.”
53:15 TW: We actually met up at Thomas Bayes’ grave, which was just cool and fun, but as we moved from there to several pubs, we were talking about magic, and how learning magic and the craft of magic that you can sort of learn some things about storytelling, and suspense and that he’s very interested in the technique of magic and doing magic tricks effectively. Not that he does a lot of them, that he thought that was actually a good skill for an analyst, which ties into my first of two little last calls, which recently it was a replay, but it’s a good one, of This American Life called The Magic Show. And just the first segment, which is Teller from Penn and Teller walking through the years that he worked on crafting this one trick and sort of the care and the craft, to me, it was kind of illustrating the love and passion for a craft, as well as, thinking through it from many different facets and since I… Literally, that was coming out very shortly after I had talked to Donal, I was intrigued by that.
54:22 TW: But then separately, ’cause I just can’t not recommend this podcast and my analytics tie-in is that Jon Levitt turned me on to it, has nothing to do with analytics, but it’s called Cocaine & Rhinestones. And it is awesome. It is about deep stories of country music from the 20th Century told by the son of David Allan Coe and Moe, from your body language, you think it’s terrible. It winds up hitting all sorts of weird cultural things. I thought it was gonna be like whatever and I’m totally hooked on it. And I highly recommend it, and I’ll stop there. But wow, Moe, you just have no faith.
55:00 MK: No, no.
55:00 TW: You’ve got so many machine-learning things you’d rather be reading.
55:03 MK: Well sometimes, sometimes. Sometimes I just go straight over [55:07] ____.
55:07 LA: I wrote it down. I’m gonna look it up.
55:09 MH: There are a plethora of wonderful podcasts out there. Sometimes we’re one of them.
55:15 MH: But this time, I think everyone will agree it was a great conversation. Lizzie, thank you so much for coming on the show.
55:24 LA: No problem.
55:25 MH: If you’ve been listening and you were thinking, “Hey, this really represents my journey” or “This does not sound like anything I have tried before,” we’d love to hear from you and you can reach us on our Facebook page or on Twitter or on the Measure Slack and we’d love to hear from you if you’ve got questions or comments about the show. I think that’s gonna wrap us up for this episode. I know I speak for my two co-hosts, Moe and Tim, when I say to everybody out there, keep learning and analyzing.
56:03 Announcer: Thanks for listening and don’t forget to join the conversation on Facebook, Twitter or Measure Slack group. We welcome your comments and questions. Visit us on the web at analyticshour.io, Facebook.com/analyticshour or @analyticshour on Twitter.
56:22 S?: So smart guys want to fit in, so they made up a term called analytic. Analytics don’t work.
56:31 LA: I’ve spent all day eyeballs deep in SQL code so apologies.
56:37 S?: This is the perfect outlet.
56:41 LA: Human interaction.
56:44 LA: It’s gonna be awkward. I’m sorry.
56:49 LA: I may or may not be drinking an old fashioned right now.
56:52 MH: That’s…
56:53 LA: See you all over Reddit/analytic.
56:56 MH: Really? We’re on Reddit?
56:58 LA: I’m kidding.
57:01 MH: All part of our strategy, let me assure you.
57:03 LA: I was trying to give a really good obscure Reddit and I just put…
57:10 MH: Hey Tim, look at my game levels. They’re through the roof.
57:16 LA: Guys, there’s so many words in this…
57:24 MH: I forced Tim to go with me to one in Las Vegas. We were very not comfortable. Yeah, he’s trying to talk less, Moe.
57:33 TW: Yeah, don’t…
57:34 MH: You’re calling him out on it.
57:36 TW: Don’t call me out.
57:36 MK: Why?
57:36 TW: Yeah.
57:38 MK: Why are you trying to talk less?
57:39 MH: He’s not. It’s just late at night. It’s super late where he is.
57:44 LA: So if you see a bearded man walk in and hand me a drink, that’s the Klein I was talking about.
57:49 MH: That’s perfectly okay. We all wish a bearded man might come in and hand us a drink.
57:56 LA: I’m supposed to feel awkward silences, right? Do I keep talking? Here I go.
58:01 MH: Is that you? Thunder?
58:03 LA: What was that? [chuckle]
58:05 MH: It’s a sign from God. He’s like “Wrap up the show”. Alright listen, this has been a really good…
58:14 MK: Tim hasn’t done last call.
58:17 MH: Tim? No, Tim lost his last call privileges when he was rude to our guest. I’m sorry.
58:21 TW: He just went all magically uncomfortable.
58:22 MH: That’s right. No. Okay. Hold on. Give you a break and then let me ask Tim.
58:30 TW: Rock flag and online learning.