NUS 77

Inspiration

Our motivation stemmed from the conviction that we could convert data into valuable insights, enabling SingLife to analyse customer behaviour. This, in turn, would empower the creation of policies that are not only better but also more tailored to the needs of their customers.

What it does

Our project endeavours to construct a sophisticated machine learning model for SingLife, predicting the likelihood of customer disengagement based on various factors, including the client's historical purchases. This endeavour aims to provide SingLife with valuable insights into the overall satisfaction levels of their customers.

How we built it

In crafting our project, we meticulously gathered diverse datasets, including customer purchase history, and employed Python scripts and advanced machine learning algorithms. Through collaborative coding sessions and iterative development, we fine-tuned our model to predict customer disengagement effectively.

Challenges we ran into

In navigating our inaugural participation in this hackathon, our team grappled with the intricacies of numerous techniques and the handling of extensive datasets. Being novices in this domain, we confronted a substantial learning curve, investing a significant portion of our time deciphering methodologies and strategies. Throughout this process, we encountered a myriad of unfamiliar error messages, necessitating extensive debugging efforts. These challenges, although formidable, provided us with an opportunity for growth and learning, fostering resilience in the face of unforeseen obstacles.

Accomplishments that we're proud of

Despite grappling with unfamiliar setbacks and unforeseen complexities, we take pride in our team's unwavering commitment to delivering our best effort during this datathon, especially considering our collective status as beginners. The invaluable experience gained in working with data analytics during this event aligns seamlessly with our aspirations for future career pursuits and academic coursework. Our team's determination and collaborative spirit ultimately culminated in the successful creation and submission of our project, a noteworthy accomplishment underscoring our commitment to the pursuit of excellence.

What we learned

We navigated the intricacies of predictive modeling, emphasizing advanced data cleaning and feature optimization. Beginning with a thorough exploration of data variances, we applied ordinal encoding for categorical variables, conducted correlation analysis, and employed a KNN imputer to intelligently handle missing values. The selection of influential features through techniques like recursive elimination culminated in a streamlined dataset. Implementing the powerful Gradient Boosting Classifier enhanced our predictive model's accuracy, leveraging its capacity for handling complex relationships. This project not only refined our data preprocessing skills but also underscored the significance of thoughtful feature selection in building robust machine learning models.

What's next for NUS 77

This is our first datathon and we hope there will be many more opportunities in school for us to learn more about data analytics and machine learning in our modules in school and through other datathon opportunities!