In June 2023, I became one of the applicants who received the FAI Institute’s Women in AI training program opportunity.  

In this article I will be sharing my journey – from application, to the completion of the capstone project. I will also give an overview of the project's role in the African technology ecosystem, shedding light on its uniqueness, strengths, and areas of growth. Then, I will address a specific community challenge, and evaluate and explain the broader implications of my project on society and industries.

Entering the program

As part of its mission to uplift the world through technology, FAI Institute responds to the issue of diversity and representation within the technology space, through such efforts as the Women in AI (WAI) training program.

The WAI training program has two stages. The first is the self-paced one, where you will learn about Artificial Intelligence (AI) and Machine Learning (ML) and select any course of your choice on the Flapmax platform. The second is the project phase where interns are divided into teams of five to execute a real-life project under the mentorship of an industry expert.

After my successful application and completion of the self-paced study, I joined a group that aimed to carry out a churn prediction project. Churn telecom prediction is used in the telecom sector to determine which customers should churn (do not continue to use a company's service) and why some customers who subscribe do not renew their subscriptions. It helps business owners decide how best to maintain their customers, and bring back those who have left, if possible. The project was empowering in that I was able to use an Intel AI platform for the first time to build the project.

My WAI Project

So far, I have graduated as a tech4dev fellow where I trained in the field of data science and artificial intelligence. I applied for the WAI training program to advance further in my career growth. This is where I utilized my skills such as using Python libraries like NumPy, pandas, and Jupyter notebook IDE. I also learned new skills using the Intel oneAPI environment and Git Bash.  

During the project stage, however, I faced the problem of power supply and network constraints. This made me even more appreciate the need for the project, especially in terms of improving the service. The experience has broadened my knowledge and concept in the field of AI and further reinforced my belief in the technology. There is no limit to which AI can offer solutions to human needs.

Through this project, network service providers in Africa will be able to provide better network service to their clients, help them make decisions for their business growth, and ensure customer satisfaction. One area of growth for this initiative involves the development of a dashboard for the network companies, which will help give them either a weekly, or monthly, rundown of their business.

Our project employed supervised learning models Logistic Regression and Random Forest Classification for predicting customer churn in the telecom sector, using the aforementioned Python 3 libraries in the process. A key aspect was comparing the performance of these models in two Python environments: a standard scikit-learn environment and an Intel oneAPI Powered Environment.

The goal was to develop predictive models for assessing customer churn in telecom, a critical metric given the cost-effectiveness of retaining existing customers in this competitive industry. This involved transforming raw data into a format suitable for modeling, including creating new columns, inputting null values, reorganizing data, handling missing values, and feature scaling.

The project required data splitting for supervised learning, with specific challenges in setting up the Python environment. We overcame these with Git Bash and addressing issues related to limited computer resources.

Our findings showed the Intel scikit-learn library variably improved modeling speed. Logistic Regression struggled with accuracy and data complexity, while Random Forest Classifier performed better due to its ensemble nature and hyperparameter optimization.

The experience I have gained so far from this project made me see the possibility and the reality of solving my identified community and country challenge. The problem of hunger and insecurity can be addressed using AI technology. Seeing that there is no limit to which AI can provide solutions, I have the idea of using an AI-based system to enhance agricultural produce and ensure food availability to the citizens. This will further reduce outbreaks of diseases, and criminal rates, reduce the poverty rate, and increase our resources.

The implications of this project on society and industries are that the local government/state government will be ready and open for partnership when they see investors willing to invest in their rural communities for their farm yield. The project will generate jobs, engage the youth, and put smiles on the faces of the farmers who are mostly rural dwellers.

Once this model becomes fully functional, it could be used as a model to reach other states in the country, repeating the same module in about 774 local areas of Nigeria.  

Conclusion: Envisioning the future of AI in community transformation  

As my internship at the culminates, I reflect on an enriching journey that stretched beyond the realms of learning and applying AI in telecom. My experience with the Women in AI program has not only enhanced my technical skills but also broadened my perspective on the potential impact of AI in addressing critical societal issues.  

The churn prediction project, while focused on the telecom sector, has been a microcosm of the larger possibilities that AI presents. It showcased how technology can be leveraged for more than just business efficiency; it can be a tool for societal betterment. The challenges I encountered – from power supply issues to network constraints – were not mere obstacles but vital learning experiences that highlighted the necessity and urgency of technological advancement in Africa.  

Looking ahead, the potential applications of AI are limitless. From tackling hunger and insecurity through AI-enhanced agriculture to fostering economic growth by generating employment opportunities in rural areas, the possibilities are boundless. The vision of employing AI to transform communities is not just a distant dream but a tangible goal. This project has reinforced my belief in the power of AI as a catalyst for positive change, particularly in underdeveloped areas.  

As we step into the future, the successful implementation of such AI models could set a precedent for other regions in Nigeria and potentially across Africa. This journey with the FAI Institute and the insights gained have not just been about personal growth but about envisioning a future where AI plays a pivotal role in community transformation. The possibilities are endless, and the journey has just begun.

‍