Training future Data Scientists - Part 5: To lead, embrace conflict
Keep your nose out the sky, keep your heart to God And keep your face to the risin' sun - Kanye West, Family Business
In the previous posts I have covered a number of items, the last being particularly focused on growth. In this post I would like to focus on what it means to grow with others. Specifically looking at how we work in the organisation and leadership of DSIDE. These lessons continue to spill over even into other work I do with the Deep Learning Indaba.
The organisation and leadership of DSIDE is basically made up of 4 core members 2 from each of the units MDS and Meraka. Given where we are all situated, physically and philosophically, the way we think of what the program means may sometimes not be the same. This eventually always rears its head at different times and with different effects. Just to go into more nitty gritty. The management of the program is not only shaping how the program looks like, which I covered in the 2nd post, but also many practical considerations like budgeting, contract management, logistics etc. It is a lot. Issa lot.
One of the earliest changes we advocated for at MDS was moving away from only targeting 3rd and 4th year students. We had a belief that targeting MSc and PhD students would increase the level of engagement and also strengthen some of the modelling work we were trying to get to. There is an appreciation that there is no right or wrong way but we ultimately pushed for this to allow MSc students and PhD students to also have an opportunity be part of DSIDE and today believe the program is better for it (MSc and PhD students are a minority of applicants and participants). The magic happens when you have all these different levels of students in the same lab working together. A similar stretch has started to happen (with a suggestion from our other organisers) on opening up a path for more business/entrepreneurship oriented students.
The largest challenge thus far has been working to sync up on what we want out of the program. A stretch I pushed for early on was the de-emphasising of creation of web apps. What we had noted was that students would focus so much energy on getting a web app to work instead of really understanding much of the analysis/modelling task in front of them. As such the likely impact of DSIDE program would be inhibited by a large focus on working out kinks in a web framework. It is understandable on the side of partners that they may be looking for a nice demo to show off what the solution is, but I tended to see it as being a distraction. Ultimately, in a full team one would have access to front end developers who would be much better suited in building these things. The 2017/2018 year saw us heavily emphasise understanding of problem and modelling as be biggest outputs to the partners for the teams. Further outputs were further grounded to allow for easier communication. In the last year we now had outputs expected being, in decreasing importance:
- Code (commented and documented)
- A report/paper draft
- A poster (Thanks to our visit to DSSG Portugal :D)
So why title the post with the word conflict. These changes small or large sometimes lead to conflict in the organising team. I could try to paper over this and ignore it, but it is a reality that has to be shared and also we should learn that we have to embrace constructive criticism and growth. Embrace the conflict and use it to also question how you do things. It is not about always being right, but having an understanding of where people thoughts come from and how to communicate your thoughts.
I tend to use the feedback from students as a measuring stick on how well we are doing and adjust some of our programs accordingly. The change to focus on modelling and insights was informed a lot from students frustration with frameworks and lack of time in modelling and being able to interact with a partner. Our partner requirements now are a bit stricter and we are very thankful to partners who showed up weekly to interact with students and get a much more agile process as a side effect.
This is fifth in a series of blog posts covering the 2017/2018 DSIDE program. We are now almost done, 2 more to go.