Inspiration
We wanted to create a project to solve Inibsa's challenge. We wanted to build a prediction model based on real-world observations instead of a complex regression model, which is hard to have an inituition for.
What it does
The system calculates a risk score for every client and every family of products. For sufficiently high risk scores it lists a warning along with the reason that the warning came through.
How we built it
We spent a consderable amount of time theory-crafting the model initially, and once we had the concept clear, we used generative LLM's to bring forward the project in a web application built with python.
Challenges we ran into
Thinking about how the model should behave without using a complicated ML model was a theoretical challenge, along with thinking the different layers of complexity that the system has to work through to bring a satisfactory result.
Accomplishments that we're proud of
We've managed to build a model to asess mostly correctly which clients are at a bigger risk of dropping without using complex ML models, which allows for comprehensive results that could be observed by-hand.
What we learned
We learnt different data extraction methods from excel databases and prediction models.
Log in or sign up for Devpost to join the conversation.