14 Apr 2020, 13:19

Meet the winners of UmojaHack #3: Hotspots Challenge

UmojaHack Africa brought more than 1000 data science students from across Africa to the Zindi platform on March 21 2020. Out of 624 data scientists from across the continent that signed up for the Hotspots Challenge, 156 made it onto the leaderboard. Only the best of the best made it to the top.

The goal of this challenge was to build a model to predict areas of land that get burned in the Democratic Republic of Congo in different seasons of the year. Figuring out the dynamics that influence where and when these fires will occur can help to better understand their effects. And predicting how these dynamics will play out in the future, under different climatic conditions, could prove extremely useful.

The winners of this challenge are: GFrost from South Africa in 1st place, Lawrence Moruye from Kenya in 2nd place and Brainiac, also from Kenya, in 3rd place.

A special thank you to the 1st and 2nd place winners for sharing some insights into how they succeeded in this challenge.

Name: Geoffrey Frost (1st place)

Zindi handle: GFrost

Github Repo

Where are you from? South Africa

Tell us a bit about yourself?

I am currently in my final year of study in Electrical and Electronic Engineering at Stellenbosch University with a main focus on telecommunication. I have a keen interest in using machine learning as a tool to solve complex engineering problems.

Tell us about the approach you took.

I started by taking a deep look at the data. This involved lots of googling and scraping through wikipedia pages to really understand what the GIS data meant. After getting to grasps with the big picture I began feature engineering, which really just involved combining existing features in creative ways based on my aforementioned research.

Next, I used fast.ai’s pre-built tabular learner function (A feed forward neural network). Once I was done with tuning my hyper parameters I created a few models each with a different random validation set, and created an ensemble of these models which then became my final solution.

What were the things that made the difference for you that you think others can learn from?

I spent a lot of time trying to understand the data and on feature engineering. Ensembling also helped a lot in reducing the models’ loss.

What are the biggest areas of opportunity you see in AI in Africa over the next few years?

AI can be applied to so many unique problems as evident in the variety of challenges in the UmojaHack. The opportunities are truly endless.

What are you looking forward to most about the Zindi community?

The vast amount of talented data scientists eager to learn from one another.

Lawrence Moruye (2nd place)

Zindi handle: Lawrence_Moruye

Github Repo

Where are you from? Kenya

Tell us a bit about yourself?

Final year student pursuing a Bachelor's degree in mathematics and computer science with a specialization in statistics and computer science.

Tell us about the approach you took.

Preprocessing: Dropped feature with unique values (ID)

Validation strategy: KFOLD repeated with 6 splits with fairly large num_rounds to prevent overfitting and have a reduced variance between the folds

Modelling: Since the hackathon was limited to 8 hours I focused with a single LGBM model because it is fairly fast and with my experience in previous Zindi competitions, LGBM has proved to do well when the dataset is fairly large (over 50,000 records)

What were the things that made the difference for you that you think others can learn from?

Feature engineering:

  • Frequency encoding with most of features
  • Aggregate features with group statistics (I used mean, std, max, min)
  • Interaction between various features

What are the biggest areas of opportunity you see in AI in Africa over the next few years?

AI is going to be the most revolutionary technology, withthe potential to solve some of the most pressing challenges that impact Africa and drive growth and development in various sectors. With AI, agriculture will be done more efficiently thus raising yields; healthcare will be better-tailored and more accessible thus improving lives; financial services will be more secure and reach more Africans who need them hence expanding accessibility and promoting development.

What are you looking forward to most about the Zindi community?

Winning and learning

This competition was hosted by Microsoft and African Bank.

What are your thoughts on our winners' feedback? Engage via the Discussion page or leave a comment on social media.