Meet the Winners of UmojaHack Africa 2022: Monthly Insurance Claim Prediction Challenge (INTERMEDIATE)
Meet the winners · 5 May 2022, 15:10

Today we meet and engage with the winners of the UmojaHack Africa 2022 Intermediate Challenge, Lawrence Moruye, Victor Olufemi (Team Solo), and Eniola Olaleye (Team Enemy of Syntax), as they share their experience and winning strategies for the Monthly Insurance Claim Prediction Challenge (INTERMEDIATE) hosted by Zindi as part of the recent UmojaHack Africa 2022 hackathon.

1st Winner: Lawrence Moruye from Senegal, Jumia Group

Please introduce yourself?

My name is Lawrence Muroye. I am a Data Scientist at Jumia Group. I graduated from Multimedia University with a Bachelor’s degree in Mathematics and Computer Science. Currently, I am finalising my Master’s degree in Machine Learning at African Institute for Mathematical Sciences (AIMS) in Senegal. I am also a Data Science Instructor.

Please explain your solution and what set your winning solution apart from others.

On the first iteration, I concatenated the two different data sets and created features from the dataset which I think was the most significant boost.

I also created the models with XGBoost (which placed me in ninth place), but I later trained my data using LightGBM (which put me first).

What do you like about Zindi?

I have learned most of my skills from Zindi by practising in exciting competitions; I have to say that Zindi has accelerated my growth in the data ecosystem. Through Zindi, I have also created unique connections with other data scientists.

Words of encouragement for others, or advice that has helped you?

Keep pushing, don’t give up. When starting, it can get frustrating after submitting. If you are not at the top, take time to read through people’s solutions, network with other data scientists, and look at the discussions made on the competition - the more competitions you participate in, the more you learn. Keep learning!

2nd Winner: Victor Olufemi, Paul Okewunmi, Oluwadunsin Fajemila (Team Solo) from Nigeria

Please introduce yourself?

My name is Victor Olufemi. A final year student taking a Bachelor’s degree in Electrical Engineering at Obafemi Awolowo University, Nigeria. During the pandemic, I did a deep dive into machine learning and data science, where I built a strong foundation. I like solving problems related to climate change and agriculture.

How did you prepare for UHA22?

I practised by going through past competitions hosted on Zindi and other platforms. I would also go through the respective solutions, which expanded my knowledge of handling different problems in the data ecosystem.

Please explain your solution and what set your winning solution apart from others.

Combining the training and policy data was a game-changer for us. We used log transformation to transform the train labels, skewed to conform to normality.

What do you like about Zindi?

I like the Zindi community - it offers a conducive space for people from different continents to collaborate, share ideas and build extraordinary relationships in the community that help you grow.

Words of encouragement for others, or advice that has helped you?

For beginners, start by writing that one line of code. Joining and participating in competitions hosted on Zindi will undoubtedly accelerate your machine learning/data science Journey.

3rd Winner: Eniola Olaleye, Saheed Azeez, Joseph Olaide (TeamEnemy of Syntax) from Nigeria

Please introduce yourself?

My name is Eniola Olaleye. A final year student completing my Bachelor’s degree in Systems Engineering, specialising in artificial intelligence at the University of Lagos. I work at Benshi.ai as a part-time data scientist.

How did you prepare for UHA22?

As a team, we gathered past notebooks and revised them. Also, since each team member is familiar with a different subject, we used this knowledge to come up with a solution. We also all worked on several different competitions on the Zindi platform.

Please explain your solution and what set your winning solution apart from others

As a team, we took time to focus on feature engineering, we created aggregated features, specifically date-time features. We used the CatBoost Regressor to build our model.

What do you like about Zindi?

Zindi helped me hone my skills, gave me leverage to practice and built my network with people around the continent.

Words of encouragement for others, or advice that has helped you?

It is essential to practice data science skills consistently. Don’t be a solo learner, try to reach out to different people.