Using Data Science in Zim Financial Institutions

Share this with your networks!

Managing risk in the financial services sector can be like playing Russian roulette. Because it is shrouded by so much uncertainty it can quite possibly cripple businesses. Data Science and machine learning can be used in a myriad of ways by Zimbabwean financial institutions.

Data science uses methods, processes, and systems to extract knowledge and insights from both structured and unstructured data sets. With the growth of the data economy, companies are able to collect more data about their users. They can collect information about demographics, behavioural patterns, attitudes and beliefs and other personal information. This allows them to perform analytics that helps uncover new and meaningful insights. Tools like Artificial intelligence and machine learning can be applied to give faster and better predictions on customer data.

There are many ways in which financial institutions can make losses. These vary from unfavourable market changes, failed operational processes, or simply credit risk from defaulted loans. It is therefore wise to leverage emerging technologies, like machine learning, to mitigate these risks.

UmojaHack Zimbabwe

A great way to come up with solutions is through hackathons, to bring together young and vibrant innovators to formulate solutions. Zimnat Microfinance sponsored a recent UmojaHack Zimbabwe hackathon, targeted at university students. The hack which was hosted by Data Science Zimbabwe, and U Lab, brought together over 30 data scientists from tertiary institutions in Zimbabwe. The innovators were provided with a sample dataset of over 12 000 loans. They were then tasked to predict which loans would not be paid off.

A participant of the UmojaHack, Cresentia Moyo, explained why using tools like machine learning, have a greater advantage when analysing large datasets over conventional methods.

“We were working on loan repayment prediction using Zimnat datasets. Unlike statistics which uses a sample of the given data, in data science we work with the whole population hence the need for bigger datasets and as a result the outcomes are usually not biased. Machine Learning makes things easy by providing algorithms that will make our predictions more accurate and also providing clear graphs for analysis”.

The solutions stemming from the hackathon were derived using tools like Catboost and Xgboost algorithms. The predictions from some the participants had accuracy scores of up to 80%. This means with enough resources, companies like Zimnat can build systems that allow them to mitigate risks and maximise their Return on Investment.

To follow the proceedings of the hackathon, follow this link.

Share this with your networks!

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.