My Data Science Nigeria 2020 Bootcamp Experience

by Moshood Yahaya

The Data Science Nigeria (DSN) bootcamp presented participants with a unique opportunity to learn, participate in global standard challenges and learn from the very best the world has to offer in the field of Data Science and Artificial Intelligence.

The journey to Data Science and AI for me has been largely a self-taught one. As a professional in the field of Risk Management, I am not really afforded the luxury of time to fully enrol in physical classes. However, my learning took a turn when I realised that I have focused more on theory and needed to take on challenges. The DSN bootcamp was introduced to me by a self-selected mentor, who wanted me to face as many challenges as possible.

The program was divided to 3 sections as stated below:
● 70 Days pre-bootcamp Learning
● Pre Bootcamp challenges on Zindi Africa
● Bootcamp Qualification Challenge on Zindi Africa
● 2020 DSN Bootcamp

The program started in July, with the 70days pre bootcamp learning. This was a rigorous, demanding and yet so interesting phase of the program. DSN set up a google classroom for the 2 different streams (Machine Learning and Deep Learning). Top notch resources were provided and relevant videos for exquisite introduction to the different topics under the relevant tracks. We are to take up quizzes on the daily lectures taken, to confirm our level of understanding and we can interact with facilitators where necessary. Although this phase is not compulsory and is limited to DSN members alone (hence, the result of the quiz doesn’t count towards final selection), the knowledge being imparted was so valuable. The fact that you get to watch unique videos from industry experts made it even more interesting. Asides the knowledge acquired, it also helps with ideas of individuals to subscribe to on youtube, for relevant solutions to problems whilst taking on challenges.

To wrap up the 70days of ML training, we were given real life challenges in the form of the prebootcamp hackathons. We worked on two hackathons namely:
1. Expresso Churn Prediction Challenge
2. The Excellent Store Challenge

The two challenges offered unique opportunities to solve real life problems that can be easily related to, using machine learning. Expresso Churn Prediction Challenge focused on customer churn prediction for Expresso. Expresso is an African telecommunications company that provides customers with airtime and mobile data bundles. The objective of this challenge is to develop a machine learning model to predict the likelihood of each Expresso customer “churning,” i.e. becoming inactive and not making any transactions for 90 days. I took a step further to deploy my model on this challenge and wrote an article on the challenges faced on the project, click here to read it.

The Excellent Store Challenge on the other hand presented an opportunity to build a predictive model and find out the profit returns on each product at a particular store, in the chain of stores owned by the Excellent Store group. This is aimed at helping the store owner to understand what type of product, market clusters and store type (location, age, size) will give more profit returns as he plans to expand to more cities in the country.

Just like the preceding phase of the program, the results of the pre-bootcamp hackathon does not determine the final selection for the hackathon, but only helps to introduce participants to real life challenges. It is believed that the most valuable way to learn data science is to take on projects and this was achieved to the fullest, particularly when you have to compete with other participants and struggle to increase your model result.

The next phase is the bootcamp qualifications challenge/hackathon. It was a keenly contested challenge on the use of Machine learning to predict customers who are likely to default loan repayment. 1095 individuals registered for the challenge, out of which only 750 made it to the leaderboard. The competition was so keen that the variance in the scores of the top 400 or 500 was only minute. It was at this point I cherished the importance of feature selection and engineering. Additionally, the choice of model for a very good result became very essential. I ended up as the 406th person on the leaderboard and this was enough to qualify me for the bootcamp proper.

At this stage, one would have thought that the excitement and drive would have been exhausted. But then came an enviable line up of programs for the bootcamp proper. Participants were spoilt with choice of classes to attend as you get to listen to and interact with the very best individuals in the world of data science. People that I have only dreamt and hoped to listen to, became readily available at no cost. Additionally, the line-up of topics was unbelievable. It was an opportunity to brush up my PowerBI skills whilst also learning a lot on Reinforcement Learning, Deep Learning, Anomaly Detection etc. My favourite session was the “AI venture funding — emerging opportunities” by Osayi Igharo. This introduced us to the very valuable discussion on monetising all that we are learning. How to seek funding and the different stages involved. This further reignited my desire to imagining areas that require AI related solutions, with possible financial potential.

The program was rounded up with awards of Mr. and Mrs. Algorithm. Certain institutions won grants from the organizers and teams with the best poster presentations on different noble solutions won awards as well. A number of DataCamp vouchers were also won by various quiz winners. Even with known awardees during and prize winners, one thing stood out, every single participant won. We won knowledge, we won unique mentorship and the desire to solve problems, this cannot be quantified in monetary terms. DSN bootcamp is unique, unrivaled and must be attended by any aspiring data scientist.