Inspiration
Every month most of us plan our monthly budgets really well but somehow few of us end up not achieving our financial targets for the month. We as humans are driven more by emotion and biases rather than data and rationality. This is where FinAI comes into the picture.
What it does
FinAI is a tool powered by advanced Artificial Intelligence models that understands your spending behaviour, compare them with your previous behaviours and analyse the goals that you have set for yourself and predicts whether you are gonna achieve those financial targets or not for a particular month. Moreover, it highlights the reasons as to why to fail to save your desired amount of money and where are you overspending.
How we built it
FinAI is written almost entirely in Python. UI is developed with the help of the Bootstrap Framework and integrated using the Flask Framework. For the AI and ML part, we have used our custom implementation of several techniques used in the field of Explainable AI (XAI)
Challenges we ran into
The main challenge here was to explain the reasons for user overspending. The algorithms that we have used in FinAI are very powerful but at the same time very difficult to explain its predictions.
Accomplishments that we're proud of
Explaining the predictions of modern advanced Machine Learning and Deep Learning algorithms is more difficult today, than ever before. We are proud of our custom solution that solves the problem of explaining the predictions of some of these complex yet super powerful algorithms.
What we learned
Because of this project we have developed a deeper understanding of predictive models. The gap between rationality and reasoning is something that has been limiting us from achieving a lot of things and having a more pragmatic and data-centric approach can help us overcome that.
What's next for FinAI
Up next we want to take it ahead and include not just personal finance but other sectors like retail, inventory management, user behaviour analysis etc.
Log in or sign up for Devpost to join the conversation.