#54 Women in Data Science (with Reshama Shaikh)

This week I'll be speaking with Reshma Sheikh about women in machine learning and data science, inclusivity and diversity more generally, and how being intentional in what you do is essential. Reshma, a freelance data scientist and statistician, is also an organizer of the meetup groups Women in Machine Learning and Data Science, otherwise known as wimlds and PI Ladies. She has organized wimlds for four years and is also a board member. We'll discuss her work at wimlds and what you, our listeners, can do to support and promote women and gender minorities in data science. We'll also delve into why women are flourishing in the R community but lagging in Python and discuss More generally how NumFocus thinks about diversity and inclusion, including their code of conduct. All this and more. Welcome to Data Framed, the weekly datacamp podcast, exploring what data science looks like on the ground for working data scientists and what problems it can solve. I'm your host, Hugo Baun Anderson. You can follow Datacamp on Twitter @Datacamp and Megobown. You can find all our episodes and show notes@datacamp.com Community Podcast this is Data Framed. Hi there, Reshma, and welcome to Data Framed.

Hello, Hugo. Thank you for inviting me.

It's such a pleasure to have you on the show. And some of our listeners may not know, but when we initially launched Data Framed, we had a panel down at Two Sigma on Sixth Avenue in New York City around Tribeca, a panel on which you appeared. So from kind of the genesis of this podcast, you've been involved in a variety of ways, right?

It was January of 2018, almost a year ago.

It was. And I actually hadn't been down there until two nights ago where I went back to present at Jarrod Lander's meetup.

How did that go? I saw that there was a meet up there with you.

It went really, really well. I gave a talk called what Data Scientists really do according to 50 data scientists. It's essentially kind of my takings from, you know, 50 hours of season one of Data Framed. And I joke, but it's true. The great thing about giving this talk is that I get to present other people's opinions and not be held accountable for them.

That's great. Was it recorded?

It was, and Jared's put it up. So it's definitely, definitely up there somewhere.

All right. I would look for it.

Yeah. It was great to launch the podcast with you, and it's great to have you on near the start of season, season two, particularly to talk about things that you're instrumental in thinking about and that you're passionate about. For example, your work at would you say Wimble ds? I'm trying to figure out how to pronounce that acronym.

Yes. For short wimbleds after saying Women in Machine Learning and Data science.

Great. So very excited to have you have you here today to talk about what you do with women in machine learning and data science and women in machine learning and data science in general. Your work with NumFocus on the code of Conduct, a blog post that took off that you wrote recently. I mean the title is as you know, why are women flourishing in the R community but lagging in Python? So I'm really excited to be talking about all of these things with you today. But before we get there, I want to find out a bit about you. So perhaps you could just start off by telling us what you're known for in the data community.

All right, so I'm known for a few things. I'm an organizer, as you mentioned, for women in machine learning data science. I'm also an organizer for PI Ladies. I'm also a board member for Women in Machine Learning Data Science. I've been an organizer for about four years now and at the time that I started organizing we had two chapters. So it's been good to see it grow and contribute to that. I also created a repository of documentation for the well known Fast AI Deep Learning Library and that's been popular. I also get workshops throughout New York City, a few online, but mainly in person and I've given a dozen in the past two years. I'm a member, as you mentioned, of the NUM Focused Diversity and Scientific Computing Committee DISC for short. And I'm known for some of my blogs.

Fantastic. So that is a wide ranging, I think resume or description of, you know, a variety of thing things you do. I'm just wondering how you, how you got involved originally. How did like what is your path or trajectory to data science initially?

Sure. So I have a, have a degree in statistics, I have a master's in statistics and I worked for a long time as a biostatistician and during when I was working there I started my MBA part time at Stern School of Business and the first couple of years were all core classes in economics, finance, organizational behavior. And then after the first half of the MBA we take elective courses and so I took a course in, I believe it was spring of 2012, it was called Data Mining for Business analytics with Professor Foster Provost and we used WEKA actually and towards the end of Our course, we were analyzing project data and he mentioned, you know, you could do that really easily in Python. And I had never heard of Python at the time, downloaded it, but wasn't sure how to use it. And so the following semester, Professor Foster Provost offered another course called Practical Data Science. It was the first time they had it at Stern in the School of Business. And that's when I started learning Python and I just. I just fell in love with Python.

Fantastic. And actually, so I was working in cell biology, thinking about, you know, data statistics, cell mechanics essentially, as well, and. But when I started thinking about working in data science in industry, it's Foster Provost's book, Data Science for Business, which is one of the first books that I ever read in the space.

Well, that funny, because at the time, at the time it was a draft copy. So we received his draft copy for free. Yeah.

Yeah. Awesome. So what happened next in your journey?

So, well, during. So during my time at Stuart, I also took other courses. Data visualization, Networks, Crowds and markets, design of apps. And so after that, you know, having a statistical background was really helpful. After that, I worked at a data science bootcamp and I acquired even more skills in data science.

Great. And I think boot camps are definitely. I mean, people ask me about the quality of boot camps and what you get out of them. I think two very interesting aspects of boot camps are actually, firstly, the network that you build and where those people end up as well, because you'll be in a class at a bootcamp, and a couple of years down the track, you'll have part of your cohort being at Facebook, Twitter, Datacamp, LinkedIn, all of these places. So that's very exciting. But as you say, the other. The other aspect is you get a broad overview of kind of all the current tools and techniques that are used in data science. So it helps you really establish the knowledge of what your toolkit may need to be.

Absolutely. You know, when I was at Stern, they didn't actually teach us about the Python libraries. We had to do a lot of coding from scratch. And there are things that are not covered in, you know, university programs, like Git, you know, at the time, not even cloud computing. So it was nice to get some of those other supplemental and complementary skills.

Absolutely, yes. Such as, as you say, cloud computing, like figuring out how to use aws.

Right, Absolutely. Yeah.

I'd love to jump in and hear about your work at Women in Machine Learning and Data Science. So, perh, you could start by kind of giving us a brief rundown of the history Of Wimbleds and the mission?

Sure. So Women in Machine Learning was founded December of 2013. So five years ago we just celebrated our fifth anniversary by Erin Liddell who's out in San Francisco and she started the first chapter there, which is the Bay Area chapter. And it was inspired by the Women in Machine Learning workshops which are co located with conferences such as Neurips, ICML and cold. But the thing about those workshops is they're only accessible to people who attend the workshops. And the goal of the Women in Machine Learning Data Science meetup groups is to have local communities where more people can participate. We are a nonprofit organization and as of Today we have 37 chapters in 17 countries with over 21,000 members.

That's amazing. And how can people get involved?

The best way is to look for a local chapter close to you. And you know, we encourage sort of people, if there's somewhere where a chapter doesn't exist, is to consider starting a chapter.

So as stated, the mission is to support and promote women and gender minorities who are practicing, studying or interested in the fields of ML and data science. How do you go about achieving this mission?

So one of the things that we do is for our speaker events, we look for women speakers. We want our community of women and other underrepresented groups to see women speaking and in a leadership position and to inspire them. And so our panel talks, we require that it's at least 50% women. We advocate and promote for conferences and we work closely with some conferences such as the O'Reilly Media and Machine Learning Conference that are really dedicated to having a diversity of speakers. And we also work with conferences that really promote a code of conduct. We have a very active slack team where women can network and ask questions about jobs and how to approach different issues at work and network. And we have all different sort of types of events such as career development panels. We recently had a terrific panel with a group of women at McKinsey that talked about sort of the challenges they face at work and how they manage their personal life with work life. We have open source brands. As you know, the rates for women being involved in open source community are quite low. And we have hands on technical workshops and examples are deep learning, Neo4J, making a bot and you know, we really want to promote a supportive and welcoming environment and also communicative environment. I think one of the reasons that the New York chapter has really done so well is that we communicate with our members and we've had a good team of volunteers that's been consistent for Close to four years.

We may get back to this, but I kind of just want to zoom in on the fact that you mentioned open source sprints and we may get back to this when discussing kind of the why women. Women are flourishing in the R community and perhaps lagging, as you say, in Python, but the open source development community I think is one in which we see a serious lack of diversity. That's an understatement. So I think, as you say, having open source sprints where you hopefully get women and underrepresented groups making pull requests as soon as possible is actually really essential work, particularly on the Python side.

Right.

I absolutely agree. In fact, I've just been spending the past week we had our second open source sprint at the end of September and we merged about that five pull requests and we had about over a dozen that are sort of outstanding and I've been working with some other people to get some of those pull requests complete. So yeah, it's a great experience. It was my first Scikit Learn sprint. I'm like, wow, there's a lot that I could have learned in the past five years.

And this is actually the first time I heard about your work at Women and Machine Learning at Data Science was when Andreas Muller, who has worked with you on these sprints, and for those who don't know Andy, he's a co maintainer and co contributor of Scikit Learn. He was running a sprint with you. Or maybe it was actually a workshop several years ago. But it was definitely then that I started hearing about your fantastic work at Wimbledias.

Oh, that's good to hear.

Yeah. In New York City. So what other types of initiatives do you think about and implement?

Other initiatives?

Yeah. So we've talked about career development and panels and open source sprints.

We've done two hackathons. They're a lot of fun. They're also a lot of work. But the hackathons and they're really well organized. There's a lot of upfront work that goes into them and we work with different communities in the city, so those are really popular too.

Fantastic. And you also have, you know, either networking events or sharing job postings or that type of stuff.

Right. We have a job board on our website where companies can post jobs and we share them. We are newsletter and also on Slack. As organizers, we really try to get to know our members and talk to them and sort of create the space that's welcoming and open. And you know, when I go to our events, I will see we tend to have about 50 to 60 people and about 90% are women. And then, you know, when I go to regular Python data science events in the New York City community, it tends to be about 80, 90% men. And it's just interesting that we are able to attract so many women to our events.

I couldn't agree more. And maybe we could. I mean, I've got a relatively ill formed question about, you know, how a lot of these aspects are, are cultural and that part of our job is to push back in terms of providing, you know, supportive and welcoming environments for everyone. And we spoke about Jared Lander and I had him on the podcast last year talking about how he's gone about building data science communities with his meetup and his conference. And actually in your post on why women are flourishing in the art community, you point out that at Jared's conference, for example, he actually provides a very supportive and welcoming environment for a lot of underrepresented groups, including women. And in terms of the speakers present, it's almost at parody in terms of gender.

It is. I mean, I have attended some of his meetups and I went to the art conference. The last one I attended was in 2016 and it is, you know, it's really relaxed and fun and welcoming. Yeah, great. He really puts on great events.

I couldn't agree more. What does 2019 look like for Wimble DS? What type of stuff you do you have going on?

We are growing, I would say of our 37 chapters, 23 were added in this past year alone. So to sustain that growth, we are looking for a sponsor. I don't know if you're aware, but our ladies are supported by the R Consortium and PI ladies is supported by psf. And what supporting means on sort of a fiscal basis is they cover the meetup dues and provide Meetup Pro accounts, which makes it easier to manage all of the chapters and communicate with them. But they also sort of provide these conduits and relationships to other organizations within the community. And women in machine learning. Data science is sort of, it's in the middle, it's not quite connected to. I don't even know if there is a machine learning organization to connect it to. But we are looking for a sponsor.

And if any of our listeners were interested in checking out the details, where could they do that?

The details about women in machine learning. Our website is our acronym, so it's wimlds.org we are very active on Twitter. Also the same handle wimlds and you can email us at info wimlds.org great.

And if people wanted to find out more information about potential sponsorship, is there a landing page for that or should they reach out?

We don't have a landing page for that, but you can send us an email to our info email address and that gets forwarded automatically to all the board members.

Yeah, perfect. So this is kind of a brief wish list of what you're working on, what else is on your wish list for WIMLDs?

So these are things that have been on my wish list for a while, and I would love to make some progress on it. And so one of them is, I mentioned the sponsors and, you know, getting a Meetup Pro account. But I would love to record our events. We have smaller chapters in, say, Michigan, in Texas, and they don't have access to all the speakers that we are so fortunate in New York City or the Bay Area or some of these large metropolitan areas. And so I would love to record our events in the same way that, you know, Jared does for the R community and to, you know, to be able to have that accessible to other, other chapters and also to sort of see what, you know, like, our Paris chapter is very active. Well, fortunately, I could understand the French, but sort of see what other people are doing, sort of having. Getting access to space. Probably the most time we spend as Meetup organizers is getting space. And it would be great to have, like, great, you know, relationships with companies where that was more accessible. And I think in the past, like even a couple of years ago, we would sort of promote conferences without really thinking as much about what their initiatives are and whether they had a code of conduct. And so we want to be really intentional about working with conferences that have a code of conduct and would really want to bring more women into open source. And ultimately all of these initiatives will go down the pipeline and bring women into open source. Machine learning, data science, artificial intelligence.

So we mentioned that listeners perhaps could get involved in sponsorship, but I'm wondering what data Frame listeners, both individuals and organizations, can do across the board and in general to support and promote women and gender minorities in data science and machine learning.

I would say the most important thing that I notice about my interaction with the community, being an organizer, is I would say be intentional. I will hear from organizations we support diversity. It's become such a hackneyed expression to really be intentional about what does it mean to support diversity and what do you do and where do you need help and. And creating, like, these inclusive women's spaces, which means, like, giving women a chance to speak. I personally would like to hear companies talk about we support pay equity, you know, and sort of, you know, seeing some data that supports that.

I think these types of statements are incredibly important because as you say, saying we support pay equity as opposed to we support diversity has a concrete outcome. Right?

Absolutely.

We'll jump right back into our interview with Reshma after a short, short segment. Now it's time for a segment called datacamp User Stories. I'm here with David Sadolsky, Founder and CEO of Boulder. David, can you tell me a bit about what you do at Boulder and why you founded it?

Hey Hugo, thanks for having me. After living and working in Manila for more than three years, I started Boulder because I wanted to rewrite the rules of outsourcing. At the time of Boulder's founding, most BPOs or business process outsourcing companies were really focused on being profit driven. And I saw that this industry could have a meaningful impact through purpose driven outsourcing. So in a nutshell, Boulder is a purpose driven outsourcing company. While we offer tailor fit solutions in customer experience, sales enablement and data management, our company purely exists to help people grow and develop at our current growth stage. My main responsibility is to empower a Boulder team to create a positive and measurable impact with our clients, company and community.

Great. So what does your day job actually look like?

My day job consists of meeting our clients, which are primarily fast growing companies all around the world, collaborating with our team members and brainstorming new ways to give back to our local communities. Other than providing financial support to local NGOs. Promoting a culture of volunteerism in our company is essential to our DNA. Last year our team has put in over a thousand hours to our impact initiatives and we supported 41 scholars globally.

So I know that you feel Manila is in a region that perhaps will be hit hardest by the data and AI revolutions. Why is that?

Outsourcing is the largest contributor to the Philippines economy. It used to be OFWs, you know, overseas Filipino workers. However, now it's driven by the BPO or business process outsourcing industry. And most of the work being done in Manila by this industry are routine or repetitive tasks. And this is primarily what the industry would call low skilled tasks. There was a study funded by the by ibpap, which is the International Business Process association of the Philippines that were really tasked with modeling what the future state of our industry would look like by 2023. And they predicted that in that study that the advancements in machine learning, artificial intelligence and other technologies will really displace up to 30% of these low skilled tasks by 2023. That same study argues that with this new wave of technology in the Philippines, we'll see an increase in medium high skilled tasks, but we'll see.

So do you agree with the study that technology will continue to replace human work?

I do. I really do believe that technology will continue to replace a lot of these low skilled tasks. And the thing that I really want to stress on is that I don't think that the local BPO industry is ready to take on these medium and high skill tasks because at that point we'll be then competing at a global scale. I believe that other countries, governments and communities are working really hard to establish a foothold in these, in these high to mid level tasks. And other markets and countries are already seeing an injection from government to improve STEM courses. And the tech community are really investing in developing their own people.

So what does this mean for the Philippines?

Let's do the math. For simplicity's sake. Let's assume there's roughly 1 million people working in the outsourcing industry in Manila. If 30% of these jobs will be displaced due to technology, that means that 300,000 people are at risk of losing their jobs. This is really why we started Boulder. We acknowledge that technology will continue to evolve the way people work. And we're really looking to ride this next wave of technology by aggressively adopting it instead of being consumed by it. Personally, I believe that we as humans want to learn and grow, but sometimes we lack access to high quality education because it's not easily accessible. It's why we've been working hard to find like minded companies to support us in our journey of helping people grow and develop.

Thanks, Dave. So now you've set the scene. Let's come back, come back later in the episode to discuss how you think about and implement upskilling your employees at Boulder and how you've used Data Camp to do so.

Thank Hugo. Looking forward to it.

After that interlude, it's time to jump back into our chat with Ray Shma Sheikh. And so you mentioned the idea of being intentional and intentionality and one example you cited was in terms of having something concrete like supporting pay equity. Is there anything else you want to say about intentionality in general or that means on a daily basis for people?

I want to point to write Speak Code. Right. Speak Code is an organization also of meetup groups and they do a national conference and they are really, I would say they've really been at the forefront of what inclusion is. And so, you know, one of the other things about inclusion is using inclusive language. So often I hear as an example, you guys, and I think it's important to be really, really careful of how we communicate with each other in the community. Because it's just. It's like the default is you guys. But really there are all sorts of different kinds of people. And there's a Discover cookbook that's put up at NumFocus about making events more inclusive to actually research that and to implement. It doesn't have to be all of the recommendations, but implement some of them and reach out if people have questions. And also, you know, collect and share data about who people are hiring and retaining and be open to sharing publicly what's working and what's not. So companies can learn from each other. So often people reach out to us and say, we want to hire more women. How can we hire more women software engineers or women data scientists or women in machine learning. We just can't find them. And I would, rather than changing the conversation from we can't find women to how about, you know, how about adjusting our environment so we can attract, hire and retain more women? And, you know, I think companies, it's important to companies to realize that people speak to each other. So when a culture is working, or if it's not working, they communicate with each other at dinner or via email or via Slack. And it's important to, you know, have a culture that people speak about positively.

Absolutely. And I do think the one flip side of intentionality is also developing awareness around what you're actually doing and what the practices you have in place actually what the downstream effects are. That's incredibly vague and abstract. I'll give one concrete example, which is the language of job listings. So I'll try to dig this up. But, you know, LinkedIn actually did several studies into how the language of job listings will even affect the ratio of applicant gender or something along those lines.

Right, that's true. And, you know, there has been research that shows, for instance, everybody knows, you know, men will. Men will apply for a job if they fit 50% of the listed requirements, whereas women will only feel comfortable applying if they. If they think that they can do 90% or more of the requirements. And, you know, knowing that that research is out there, companies could adjust their job postings if they want to, you know, attract more women.

Yeah.

You mentioned NumFocus and the diversity and Inclusion in Scientific Computing Committee, otherwise known as a disc that you're on. And I'm wondering how you think about diversity and inclusion at num. Focus as an organization and what type of initiatives you're involved in there.

I think they're really, really cognizant about the importance of it and how to sort of move that initiative forward. They've been fortunate to receive a grant from the Moore foundation, and they have all sorts of different initiatives. And so one of the things that has been their project is to have a comprehensive code of conduct, which I, you know, it's interesting. I thought, oh, I'm sure somebody else has done it, because I was researching it for the non focus committee, and I was surprised to find out that there is actually not too much out there in terms of comprehensive codes of conduct and, you know, defining acceptable and unacceptable behavior and how to report. And so that's, I mean, that's really critical. And I think that other organizations, you know, outside of non Focus, outside of their projects, can. And all of that information is publicly available so they can access it and reference it and use it to make their communities more inclusive and diverse.

Great. And so maybe you can speak to the idea of a code of conduct in general and what it provides for a community at large.

A code of conduct is so important. When I started organizing four years ago, we didn't have a code of conduct. And there were sort of these behaviors that I would observe at our meetup events, and I wasn't quite sure how to address them. And then codes of conduct came along and then it made it so much easier to, to say, you know, that behavior is unacceptable because it violates our code of conduct. And now sort of, you know, I've moved forward to sharing the code of conduct before the event begins. So people know what behavior is expected, and that's been like a tremendous help. And I think that for people who attend events, knowing that a code of conduct, that this is important to the organization, it's, you know, it makes them feel comfortable in terms of, of attending events and knowing how the culture will be.

And once again, this comes back to being intentional, making things precise and making things concrete. So, for example, this idea of we value diversity, it isn't clear, you know, how you would implement anything around that. Whereas writing a code of conduct puts certain precise steps in and a precise code down.

Right? Absolutely. And so actually, this is a side thing I wanted to, you know, mention about. I wanted to say something about the diversity. As an example, we've had companies in the city in, know, one major tech company that's reached out to us that said, you know, we want to hire more women. We can't find them. And so we talked about an event and after, you know, many emails and conversations, they said, you know, we just don't have the budget for it. And this is, this is a tech company that provides three meals a day to their employees and they've said that they want to hire more women, but then they say they don't have the budget for it.

Yeah, that seems hypocritical to say the least. Or willing to say something, to talk the talk but not walk the walk, so to speak.

Speak.

Right.

And speaking of codes of conduct, you have a blog post on codes of conduct in general, right?

Yes, I do, yes.

And we'll link to that in the show notes. But I think it's wonderful. So I thought maybe you could say a few words about it here on the show.

Sure. So you know that code of conduct, as we were doing the research for the non focused DISC committee, it was published and because I knew that there weren't a lot of comprehensive codes of conduct out there, I wanted to write about it so that other organizations would know that it existed and that if could help create a welcoming and inclusive and professional environment for them. And I really hope that people will take some time to read this blog because it's not one that has received that much attention. But it's probably one of the blogs that I've written that's the most important to me and I think it's because it has the potential to really help. I'm going to focus on data science and STEM associations, but really it's applicable to a wide field, a lot of different other fields of work. But once they going from, oh, you know, we, we're a welcoming committee community. This is actually more concrete actions like having a code of conduct, it can prevent inappropriate behavior in the community, it encourages professional inclusionary behavior and it provides a safe avenue for communicating violations. And I think that, you know, I reviewed about 10 or 12 organizations that serve the data science community. So you know, in statistics, computer science, machine learning, physics, economics or engineering, and all of these organizations, aside from say numfocus and the American Statistical association, have very brief codes of conduct without too much information and membership. Total membership in these organizations is 400,000. So if this information gets out to these STEM communities, you know, it really has the potential to have, you know, a really like favorable impact on inclusivity and professionalism in the entire STEM ecosystem and also for generations, you know, that are coming.

As I said, we'll include the link in the show notes and if any listeners have any feedback or any comments about it, please get in touch with both of us as well on Twitter or otherwise. So going back to NumFocus and disc, we've talked about the code of conduct, but there are other initiatives such as the Unconference and the Discover Cookbook and other things. Right.

In about a year ago, In November of 2017, there was a disc on conference. And so about 45 people came from the US a few in Europe, a few in South America. And we worked on different committees. So some of those committees, they finalized or worked on the DISC Events Guide on how to make conferences and events more inclusive. There's a directory of organizations that serve underrepresented groups. There was a committee on getting started with open source, communicating Feedback anonymously and diversity metrics. So there were all sorts of initiatives and it really is a great organization to get involved in. And. And so many things that I've learned through the DISC committee and being at the Disc Gone conference have helped me in my work with my meetup groups as well as in open source.

Great.

And what's the cookbook?

Ah, the cookbook. It's a list of sort of suggestions that can make conferences and events more inclusive.

So, Reshma, you worked on the Cookbook with several people at NumFocus, right?

So the cookbook is. It's a resource that provides sort of actionable, actionable items that communities can do to make their events and conferences more inclusive. And so, you know, not everything is required, but it's sort of separated into high impact and effort, whether high impact, high effort, low impact, low effort. And they really give, you know, helpful suggestions about, you know, selecting venues and food, you know, making food options accessible to people that have different dietary restrictions, providing child care grants and scholarships, how to select participants, event registration, you know, having sort of the name tags with pronouns. There's, you know, there's just all sorts of options.

We'll jump right back into our interview with Reshma Sheikh After a short segment, I'm back here with David Sadolski, founder and CEO of Boulder. Dave, knowing that Manila is in a region that perhaps will be hit hardest by the data and AI revolutions, how do you think about and implement upskilling your team members at Boulder?

The reason why we exist is to help people grow, develop. It's not to be the biggest outsourcing company or be the most profitable. And our definition of success at our company is empowering our community and our team members, which is why it's important for us to seek and work with those who believe in our vision and are lifelong learners. Just like we do with our clients, we don't assume that one size fits all for personal development. When working with our team, instead of pushing people down one path, we seek to understand their own personal goals and passions and what they want to pursue. Pursue. Lastly, we encourage them to dream and dream big.

So how much demand for data skills do you see from your employees?

When I was looking at the numbers, I was really surprised about this. I knew our team members working on data related tasks wanted this, but I was stoked when I saw that nearly 50% of our entire company have completed Datacamp's classes.

That's awesome. So how has Data Camp helped you achieve your mission to help people grow and develop?

Datacamp was instrumental in bringing once hard to access classrooms right into our office. Our team members are able to enroll in Data Camp courses, work on problem sets together, and also collaborate on practical applications of their learning. Right away, we've set an aggressive goal of promoting roughly 20% of our team and our learning and development organization is working on making this a success. And with Data Camp, this goal becomes a little less daunting and more attainable.

Can you tell us about any particular success stories at Boulder with respect to developing data skills?

Yeah. So John, who's one of our Data Explorers, was incredibly interesting. However, as a company, we really failed at identifying courses for him. Now with Data Camp, it has allowed him to self study Python and SQL, which in turn allowed him to handle new projects at our company. I've been working with John since I moved to Manila and I'm proud to see how the complexity of his projects has increased dramatically. And the most beautiful thing about this is that because of his experience with these new technologies, we're able to work with him, work with our own client base, and offer more complex data solutions. And you know, upskilling our team isn't only good for us as a company, but it's also a win for our team members as their own income and spending power increases.

And what data skills in particular have you seen the most demand for and what's most important for the work that you do at Boulder?

The most demand we've seen in specific programming skills is like SQL r Python. However, for the work we're starting to handle, I would say statistics and data visualization are the most important.

Yeah, we see the same across a lot of different verticals. So Dave, thanks so much for coming on the podcast to share a portion of Boulder's data journey.

For sure, Hugo. It was fun.

Time to get straight Back into our chat with Reshma. So now I really want to jump in and talk about your recent blog post. Why are Women flourishing in the R community but lagging in Python? And for the listener, I just wanted to give a bit of background into this blog post. Reshma and I, couple of months ago, we're chatting on, on a call about what a podcast conversation could look like around these issues. And I'd been at one of my favorite conferences in, in the world is scipy, which happens in Austin, Texas every year. And I was at a meeting at Scipy in, in July, in 2018 and it, it came up how to think more about inclusivity and diversity at SciPy, which is a big challenge. And I kind of off the cuff mentioned that anecdotally at least the R community and R conferences seem to be doing a better job at inclusivity and diversity in general. And so there was a big conversation around that. And I actually asked Reshma about this and she said, that's a great question. Or you said, reshma, that's a great question. I was just speaking to the listeners for a second and decided then to go and look into it and publish this blog post, which really, really took off.

Right, right. You know, it was sort of a comment at the end and I had some ideas about why and I thought, well, let me do the research and put it together in a blog post for our conversation today.

And so maybe you can tell us some of the key takeaways from your research and the post and then we can discuss the responses.

Sure. I would say, you know, the key takeaway from all of that research and you know, one of the readers, he really said it best, he wrote the key takeaway here seems, and this is actually Dylan Niederhut, he said the key takeaways here seems to be that R has top down institutional support for inclusion where efforts in Python are less supported and less concerned with each other. I think that pretty much sums it up.

I think that's a very concise and very pertinent takeaway. What does the data actually tell us or show us?

This is something that there's all sorts of metrics. For instance, it's not something that's easily measurable like taking a blood test and knowing exactly what the blood cell count is. And so the question is, well, how do you define inclusivity? And so one of the ways is look at it is look at it open source contributions, the percent that are members of the communities, whether it's the general communities or the aspect of say women in this programming language community and you know, the percent speakers and how many people submit calls or proposals, how many women are involved in different places in the community.

And so what does the breakdown look like between Python and R?

The open source contributions for I would say for R are just so much higher. So one of the things that I did discover in my research is that there's not a lot of data out there about Python. It's very limited data and I think that collecting and reporting data is, is indicative of how much the community is where diversity inclusivity is a priority for the community. And I think that Python is a bit lagging instead of having, maybe they have the data, but it's not really, maybe it's not published or so. But in terms of open source, you know, for R, it's four times the number of package authors that are women versus in Python.

Okay, so this idea that R, one of the takeaways is that R has top down institutional support for inclusion. What does that look like in terms of the community? Where does this institutional support come from?

Well, I mean it goes, it is all the way from the top. So the R community has the R Consortium which, you know, manages the, I would say, I mean you, I know you're very involved in the R community, so, you know, I guess the R Consortium is sort of the equivalent of the Python Software Foundation. Would you agree with that?

Yeah, absolutely. I think it's a nice analog.

Okay. All right. And so the R Consortium has, you know, announced that R Leanings is a top level project for them. And what does that mean again? You know, what do these words mean? Top level project? You know, it's important to them. What they do is that they provide a budget cycle for them that I guess it used to be renewed annually, but they have a budget cycle now that's set for three years and they have given, I mean this is really impressive is that they've given our ladies a voting seat on their infrastructural steering committee of the R Consortium. And comparatively speaking, for the Python Software foundation, they support fiscally, like they pay the meetup dues for PI Ladies, but it's only the PI ladies who are part of their Meetup Pro account. They haven't really converted them and they don't really, you know, they're sort of part of Python Software foundation, but there's no, there's no structure there.

We've discussed how, you know, a lot of these aspects are cultural and historical as well and I actually have no idea Whether what I'm about to ask is true or not, but at least my intuition and anecdotally knows that the R community historically has been a very statistical, academic, pedagogical community, which already has a lot more women, for example, involved. Python, historically and culturally comes from a software dev background, which doesn't. So could these two factors play an important role?

I really think yes, I would say they definitely play a role. But there is something that I do want to point out is that both Python and R were released in the early 90s. So Python was released in 1990 and R was released in 1993. And so the women's communities were founded about 20 years later. And I think that's interesting, both of them, for Python and R, that they were founded 20 years later. And I think that whatever the educational background is, they both communities felt the need to create these women's groups to advance them in the respective programming languages. So that's really important to consider.

I agree. And the other question that springs to mind is the R community is actually a lot more of a community than the Python community. I actually even, you know, my Spidey sense, or whatever you want to want to call it, arcs up when people talk about the Python community. Because my question is you're talking about scientific Python community, which is actually a very different structure than the Python Software Foundation. And so really the point is that scipy is very different to pycon, right?

Yes. I mean, I have not attended either of them, but I know that I think one of the things about Python is there are. It's used so expansively that there are all different communities, but I think that can also be an asset for the community too. It doesn't have to be hindrance.

Agreed. So what would you like to see the Python community do in the coming years?

I would like to see a. I would like to see more collaboration between them, communication between them. They can sort of. It's probably okay for them to retain their, you know, their identity and their structure and all of that, but they can communicate with each other and, you know, and sort of like merge some of their initiatives because every, you know, everybody. This is the analogy I would say I would use for the Python community versus the R community. You know, everyone is committed to diversity. Everybody wants. Wants inclusivity and they're all working hard and they all have initiatives. But when I look at the Python community, it's like each sort of group is in its own individual kayak and they're rowing hard. Everyone's looking at the same direction and they're moving there and they're all working hard. But the R community has sort of invested in this really large boat and they can move to their goal a lot faster with a lot less rowers to get where they want to go.

Right. And I think what you're speaking to there is that everyone is on board the boat. Right. That speaks once again to it being a really having a strong sense of community there at the top as well.

Yes, yes. I think it's more of a structural issue.

Right.

We're going to wrap up in a minute, but I've got a couple of final questions and one is something I ask a lot of my guests which isn't related necessarily to the rest of the conversation because we haven't talked about data science a lot. I mean, about the technical aspects of data science. So I'm just wondering, what's one of your favorite data sciencey techniques or methodologies?

My favorite one, I would say, is data visualization. I took a course when I was at Stern with Kristin Sasolsky and she teaches really a terrific data visualization course. And I realized after all the years of being a statistician that I was not presenting my data as optimally as I could be. And so visualization, I would say, is key for sure.

And my final question is, do you have a final call to action for our listeners out there?

Yes, I do. And I would say that if you're interested in starting a chapter for women in machine learning and data science, feel free to reach out to us. Our email is infowimlds.org we have a starter kit and we are happy to answer any questions. I would say I really would love if the blog I wrote the article about code of conduct for Neurips and other STEM organizations. It's really critical and helpful information. Non Focus will be having nominations for their committees beginning in March, and so I imagine they'll post it in February or so. And I would say get involved in the various committees that can really help with wherever people are working, whether it's at a company or working in academia or just running their own community. And sponsor if you're interested in sponsoring women in machine learning, if you are interested in sponsoring women in machine learning and data science, also reach out to us@infoiml.

So those are all fantastic final calls to action. And in terms of reading your blog posts, I'll include them in the show notes and I'll urge everyone when reading them to share them as loudly and widely as possible. In particular on Twitter. I think, as we've seen, Twitter is actually an incredible space to have these conversations, at least initially. And there is a critical mass of people really willing to engage on Twitter about this as well, Right?

Absolutely. I love Twitter. I mean, I think it's a way to. However it's designed, it's a way to connect with people that I would not normally know, either meeting them in person or even via some of the other social media platforms out there.

Yeah. And even on these issues, which are particularly sensitive and where on social media you generally, in a lot of other disciplines expect a lot of trolls to come out, a lot of these conversations stay amazingly civilized on Twitter as well, and constructive, I find.

I absolutely agree. I think that's because, you know, people are very committed and are interested in advancing the issues and making the community better for everyone. So I agree.

Agreed. Reshma, it's been such a pleasure having you on the show.

You too. Thank you, Hugo for the invite.

Thank you. Thanks for joining the conversation with Reshma about women in machine learning and data science, inclusivity and diversity more generally, and being intentional in what you do. We saw how important top down institutional support for inclusivity and diversity is with the difference between R and Python communities providing an illustrative, inclusive example. Reshma stressed the importance of concrete statements and actions to these ends. One example being that R Ladies is a top level project for the R Consortium and that the consortium provides a budget cycle for our ladies and they've given our ladies a voting seat on their infrastructural steering committee. We also saw how important specific statements such as we support pay equity are for companies and seeing data that supports this. Reshma also spoke to the value of codes of conduct with specific references to the one she was involved in developing at NumFocus. Codes of conduct promote being intentional in action and decision making by explicitly stating what is acceptable and what isn't. Next week I'll be speaking with Chris Albin, data scientist at Devoted Health, where he uses data science and machine learning to help fix America's healthcare system. Chris is also doing a lot of hiring at Devoted currently and he'll join me on the show to talk about how to get your first data science. You may know Chris as co host of the podcast Partially Derivative from his educational resources such as his blog and machine learning flashcards, or as one of the funniest data scientists on Twitter, although we're all still way behind. Vicky Boykas Hi Vicky, I'm your host Hugo Bown Anderson. You can follow Datacamp on Twitter @Datacamp and me Hugo Bowne. You can find all our episodes and show notes@datacamp.com community podcast.