Demo Day: AI_Saturday_Abuja (2nd Cohort)

“What we call the beginning is often the end and to make an end is to make a beginning. The end is where we start from.” T.S Elliot

The journey of the second cohort of AI Saturday Abuja, which started early this year, has finally come to an end today (June 8, 2019) amidst funfairs. The positive energy that engulfed the atmosphere of Ventures Park, the venue of the demo day, was infectious. Sixteen Saturdays completed already? Yes!!! How time flies. The good news is, the end of the second cohort is the incubation and birth of the third. The second cohort, just like the first, had attracted participants from various fields and disciplines (Law, GIS, Economics, Electronics, Mechanical etc) not exclusively for Computer Science. Yes, you heard me right, with basic knowledge of microcomputer utilization, you qualify to be part of AI Saturday where in beginners’ class, you learn concepts like introduction to Python, Jupyter notebook, Git, Data Visualization, Regular Expression, ML algorithms etc. The intermediate class did more of Deep Learning. Top-down learning approach using Jeremy Howard’s Fastai technique was adopted in some of the intermediate classes. I’ll give eye-witness account of everything I enjoyed, being among the privileged ones who attended the demo day (all thanks to Chijioke Eze, the friend who introduced me to the AI6 Abuja and encouraged me to continue when I wanted to quit after my first class). Fasten your seatbelt and stay glued till the end.

The day started with participants arriving before 9am. Presenters of personal and group projects took advantage of their early arrival to go through their slides and test run their codes. You can read excitement, with a tint of nervousness, on the faces of even the most confident, which was further noticeable when the coordinator of AISaturday Abuja, Simon Ubi, announced the commencement of the demo day.

Preliminary discussions centred around the plight of human traffic controllers who carry out their task under the scorching sun, with suggestions on how AI can leverage computer vision to improve the traffic control system. The day kicked off with presentation of personal projects; group projects were presented thereafter.

The first presenter was I.K Emmanuel, who developed a Loan Grant Predicting app. The highlight of his presentation was the deployment of his model using flask. It was an eye-opener for everyone in the beginners’ class considering we were yet to cover how to deploy models. It was a big motivation and I remember telling myself to include flask on the list of things I hope to learn before the next cohort.

Next came Sumayyah Musa, who presented a ML project on Relative Humidity Prediction, motivated by a Cotton Manufacturing Company. Weather readings were collected over three years, with all measurements taken in the morning except a feature. The data was used to train, compare and contrast two ML models (Logistic Regression and Random Forest Classifiers) to classify the target as either Low or High relative humidity. Random Forest was concluded to be the most suitable.

An interlude was observed after the second presentation. Simon Ubi asked if the house can work on a project that involve flood prediction. The house agreed that it was feasible, given the right dataset. He also pointed out that AISaturday is not only about learning how to implement Machine Learning algorithms, it is also a place that provides enabling environment for AI researchers. I couldn’t feel more at home after hearing this, considering I intend to pursue academic research in AI.

The third individual project presentation was by Tosin, who presented a project on Document OCR Classifier. His motivation came from the hurdles he faced when he tried to obtain his certificates from his past institutions of learning.

Around 12.45pm, Tunde, one of the senior blocks of the cohort, moved motion for a 5-minute break, before the commencement of group presentations. Simon took the opportunity to announce the presence of Dr. B.K Mustapha, an associate of AISaturdayAbuja.

Group presentations commenced immediately we resumed from the break. The projects were past Kaggle competitions. Group 1’s project was on Santander Customer Transaction Prediction — a binary classification project that targets whether a customer will make a transaction or not. Group 2’s project was on predicting future sales for a shop based in Russia. Group 3’s project was on New York City Taxi Fare prediction, which was deployed on a web app. General appraisal followed after the group presentation:

Tunde’s observation on project deployment is the possibility of making the deployed system have users select data from a list by integrating plugins with specialized libraries, instead of having to manually enter the values for the fields.

Dr. B.K appraised the cohort and above all, the demos. He gave exposition on the benefits of the demo day from an academic and research perspective. He encouraged us to look around us for problems that can be solved using even simple ML algorithms.

Simon Ubi advised all groups to leverage their collaboration to improve beyond the cohort, also advising everyone to use the break until the next cohort to intensify learning.

Concluding the demo day, Simon Ubi, commended the outstanding performance of two participants of the cohort, Emelda and I.K Emmanuel. The house celebrated them as they were announced as part of the Instructors for the next cohort. The recognition was indeed well deserved, as their devotion and commitment were glaring to all. Simon also anchored deliberations about the forthcoming Hackathon, which was slated to usher in the next cohort and the proposed hiking to be undertaken by participants will bring AI Saturday Abuja’s second cohort to a close. I intend to participate and enjoy my first hiking ever. Did I tell you we planned to rock our super dope AISaturday Abuja customized T-shirt? Keep your fingers crossed for more stories about the adventures of AI6Abuja.