If you’ve been paying any attention to the technology used by Fintech companies around the world, you would’ve undoubtedly heard of the meteoric rise of the Python programming language. Originally popular with academics, Python’s easy-to-learn syntax, expressive power and ever-growing ecosystem has propelled the language to enormous popularity. Currently, Python is considered the first or second most popular programming language in the world. And is likely to remain popular for some time. It’s no surprise the future of the credit modelling industry will be written in Python.

How Python ate the world.

Python has three key qualities that led to its superstar status: an easy-to-learn syntax, an active opensource community and a massive third-party ecosystem.

If you ask any Python practitioner what aspect of the language they love most,  many will point to the language’s expressive, simple syntax that’s easy to read and understand. This is no accident — one of the guiding principles of Python is that there should be one — and preferably only one — obvious way of doing something. By adopting this approach, Python remains easy to learn for beginner programmers, and a breeze to adopt if you’re coming from a commercial language MATLAB or SAS. Our anecdotal experiences mirror this claim: we’ve seen lots of success with our Python training courses for credit modellers, many of which are new to the language. As a bonus, Python’s popularity also makes finding help online surprisingly easy.

The sometimes-overlooked attribute of Python, particularly if you’re used to commercial tools, is that it’s a 100% open source project. For years now, the evidence has been accumulating: large open-source projects almost invariably outcompete and outlast commercial offerings once they’ve gained enough traction. The reason is simple: open source projects can benefit from a large number of contributors, can constantly be improved, and can continue to exist through the community, even if the original authors disappear. Python is also a free open source project, which means core developers get more exposure to user feedback than would otherwise be possible commercially. This suggests Python will continue getting better and better with each passing release. This community is precisely the reason why Python has steadily evolved into one of the most loved languages in the world, and will likely continue to do so long into the future.

Finally, one can hardly talk about Python without mentioning the downright vast, bustling ecosystem of third-party packages. Numerical computation? Check. Machine learning? Check. Data visualization? Check. Web frameworks? Check. It’s all there. Python’s package index, PyPi, has over 345,000 packages available for immediate installation, for free, for any niche imaginable. Wouldn’t it be wonderful to leverage this power to build better credit models?

Credit modelling is next on the menu

Credit and credit models are playing an increasingly central role in financial institutions. As time went on, credit modellers and decision scientists have been expanding their skillsets to reflect the increased importance of their role in the day-to-day of the business.  Unfortunately, many credit modellers are often forced to rely on one-size-fits-all credit solutions. This brings its own set of problems: datasets are too large, solutions don’t have the necessary features, and you’re limited by the power of your local machine. As part of our work, we speak to many clients and the financial industry, and we often hear the same things over and over again from those leading credit departments: decision scientists have started learning Python themselves, specifically to work around the limitations they face with their current modelling software. Furthermore, given the ever-increasing overlap between machine learning and credit, it’s no surprise credit modellers have shown a particular interest. Guess which programming language is the dominant force in the domain of ML/AI? That’s right, Python. As modelling and statistical knowledge and technogloy improves, more advanced techniques will continue to be applied in the credit industry. But they’ll become available in Python first, thanks to the large community. As far as we’re concerned, the future of finance lies in Python.

Enter Praexia

Today, you have a rapidly-growing credit industry full of smart, forward-thinking people that are frustrated with the tools at their disposal,  forced to build hacky workarounds in a programming language they’d rather just be using outright. The thing is, credit modellers are fantastic at credit modelling, but they aren’t experts at developing scalable, reliable software. That’s why we partnered with our clients and decided to build Praexia.

Praexia is designed to empower credit modellers by providing not only a turn-key solution that makes common credit modelling tasks like variable selection and binning easy, but to give credit modellers unparalleled access to the entire Python ecosystem. Praexia also runs entirely in the cloud, as part your infrastructure or as a managed service. This means no more “but it works on my machine” or “the data doesn’t fit into memory” problems. In the cloud, we can scale Praexia to suit your computational requirements. Furthermore, Praexia is built on top of the world’s most successful, battle-tested open source software itself, to provide a reliable, feature-rich experience. We truly believe that Python will become the future of credit modelling, and Praexia is proof of that. And we’re also not alone in this belief either: we’re already trusted by large customers such as Capitec, Matogen Applied Insight and Lulalend, who have all seen tremendous benefit from adopting Praexia and Python in their credit modelling process.

Will Python continue to eat the world? Undoubtedly. A proven track record, a large ecosystem, and a growing support base in form of credit modellers and decision scientists all but guarantees this will happen. The question now isn’t if, but how will credit modellers and financial institutions leverage the Pythonic revolution. We believe Praexia to be one possible answer. After all, the future of the credit modelling industry will be written in Python.

If you’d like a glimpse into what the future could look like, take a look our features, read the technical documentation, or get in touch.

Recent Posts