Elena Mwangi: Knowing when to stop.

5 min readOct 25, 2020

By Mwaniki Nyaga, KamiLimu Management Committee Member

I first met Elena Mwangi when we were both KamiLimu 3.0 mentees working on a group innovation project. Looking back, what particularly stood out for me was her sarcasm, morbid jokes and the graceful way she would speak while delivering a personal speech or explaining our group’s pitch deck to judges.

Almost three years later, I sat her down for an interview via Zoom to understand what drives her as a person, her fascination for data and the goals and achievements she has made as a peer mentor in the KamiLimu Data Science Track.

Elena is part of the group of mentees being mentored by Jacklyne Betty, a Machine Learning Data Scientist and the Track’s Professional Mentor.

Elena Mwangi, a student at Moringa School, Cohort 3.0 Mentee & Cohort 5.0 Peer Mentor.

Who is Elena?

Elena is a student who studies Data Science at Moringa School. She spends most of her days seated in front of a computer screen trying to figure out how to solve a particular problem. On some evenings, she mentors students while on others, she’s a volunteer in a Data Science community called After Work.

Why did you choose Data Science?

I’m the person who has 48 tabs open on their browser because I’ve gone down a rabbit hole. I go back and forth between a book I’m reading and a dictionary when I don’t know the meaning of a word I’ve read. I have to know. I think that’s how, in hindsight when I think about it, I ended up in this field of data science because you always have to answer the what-ifs, have hypotheses to test and find answers to. And I think, at its core, that is what is so exciting to me about Data Science: the idea that you have these numbers from a bird’s eye view but when you dig deeper and ask questions about your data, you discover patterns. Basically, the whole journey involved in gaining insights from data is innate to who I am.

What was your mentor’s approach to the personal projects?

Jacklyne had a calendar with 30-minute slots that one could book to catch up and give her a progress update. I felt like my meetings with her were like talking to a fellow geek. I looked forward to them because she was so good at giving input and feedback and you could see that she was just as invested. Also Mwaniki, I’ve never seen someone who responds to emails so fast! She was so prompt in answering the questions I had. I think having someone as interested in the project, as personable and as friendly, pushed me to work even harder, especially when I didn’t feel like doing so.

What was your project about?

Predicting track popularity from Spotify data based on factors like the track’s instrumentality, acoustics, loudness, energy and danceability. I wanted to see whether I could cluster music to find similarities in genres, for example, is country music loud or more danceable? To answer questions like, how do you define a genre and can one find emerging genres? At its core, it was a prediction and clustering task.

What did you learn going through the project?

How essential it is to not only work with the data one has but to also search for more complementary data. In my case, I sought to see whether artists who are considered popular on places like Billboard’s Top 100 or Rolling Stone music charts have higher popularity scores on a streaming platform like Spotify. Indeed, my hypothesis was proven true; their tracks were having a higher average in popularity. I also learnt to continually experiment. The idea of always asking more questions about my data surprised me with how much more I could actually find and learn from it. Finally, I learnt to know when to stop. When to make those close calls on whether the work product is good enough to pass on to management and being comfortable with that.

How was your experience in terms of your fellow mentees’ approaches?

The good thing with having different kinds of people in our team was getting to see a large spectrum of use cases where Data Science can be applied. We had people such as Koi doing the track to primarily get to know what Data Science is about, to Jeff who was looking to build a business viable product, to Ashtone classifying medical imagery, to Derrick collecting his own data. It was good to see the enthusiasm stick until the end — because a majority of us presented our findings. Seeing other people’s projects was also a source of inspiration and a challenge to me; it led me to ask myself what more I could do, what could I build in my next project? And being a peer mentor, I was often the first person to be approached when people had a problem, and even though I may not have had the answer, that prompted me to do my own research and share my findings with them while also learning in the process.

What could have been done differently to enhance your experience?

(Pause) I literally have nothing to say on this. I came out with a project that I really loved working on and I had a wonderful mentor. She went above and beyond in offering help, and for that I am grateful.

What did you achieve and was this in line with your hopes?

In the beginning, my objective was to work on a project from start to end. What I hoped for was someone I could ask my many questions, someone who, when I come across a challenge, I can be secure in their help. I got that and much more. In the process, I learnt practical tips regarding working in a company. Jacklyne would tell me how it’s important to know the reason behind what one’s doing because with data it’s very easy to fall into a trap of throwing algorithms onto data without knowing the assumptions behind said algorithms and what biases you might be introducing. Jacklyne also showed me that it’s okay to be a nerd; to know why it is I did something and what led me to make that decision. I was glad to get into the mindset of a data scientist working in an actual company and how that would look like.

What next for you and the group?

We’ve expressed our interests and what we’d like to work on for the group project. I’m currently really into image classification, especially regarding medical imaging. Recently, I discovered that eye care is not that accessible in Kenya, where one finds a large occurrence of treatable cataracts. Therefore I’ve started reading a paper on the use of image classification to detect cataracts. That’s one of the things I would like to do. There was also another case that piqued my interest. It involved the detection of kyphosis using x-ray imaging. Kyphosis a spinal deformity similar to scoliosis, which I actually have. The only reason I knew I had it is because we had to go to a specialised hospital. The doctors in this area are so few and we knew of their existence only through numerous referrals. I hope I get people who are as interested in embarking on such a project as I am.




KamiLimu is a free 8-month structured mentorship program that seeks to augment classroom learning for tech-aligned students at Kenyan universities.