Examples of Fintech Projects

Students assisted an accounting firm to explore whether they can build machine learning models to ensure accurate and fair holiday pay while improving overall efficiency. The Holidays Act 2003 in New Zealand poses significant challenges for businesses, leading to potential payment discrepancies. Noncompliance with the Act can result in substantial costs, penalties, and legal consequences.

Leveraging client-provided data, the students undertook data cleaning and preparation, subsequently training machine learning models to enhance the precision of holiday entitlement estimates. These models demonstrated superior accuracy in unseen and new datasets, compared to the client's existing methods. Further, cutting-edge analytical tools and explainable AI were employed to identify key features contributing to over- and under-payments, resulting in a set of improvements proposed for the client's current process.

The project involved students partnering with an IT service company to optimise its operations by analysing internal data. The goal was to enhance resource allocation, improve project timelines and increase profitability. The team mined insights from company’s internal data, spotlighting automation needs, job profitability, timelines, and team organisations based on observed work patterns.

Overall, the project has yielded valuable insights that foster data-driven decision-making, enhance resource allocation, and lay a strong foundation for future automation and AI initiatives.

This project focused on addressing the key concerns of insurance companies regarding the risk profiles of their clients.

The project tasked the student team with utilizing data provided by the insurance firm to achieve several objectives. Firstly, they were required to develop risk profiles for both existing and prospective customers.

Secondly, the project entailed developing claims trend profiles using existing claims data. These profiles enabled insurance underwriters to establish risk appetites based on claims trends, considering the nature of the claims, premises location, and property characteristics.

To accomplish these objectives, the student team developed several machine learning models, individually and in combination, for classification and further analysis purposes.

Pension Fund Advisors face a surge in online applications for advice, leading to capacity constraints for human Consultants. To prioritize consultant time, a predictive system using several machine learning models were developed by students to classify clients with high conversion probabilities. By categorizing applicants based on conversion probabilities and evaluating the impact of consultant interaction, resources could be allocated more efficiently. This solution aimed to streamline the process, ensuring that consultants focus on applicants who are more likely to convert, while allowing some applicants to proceed through an automated process.