Sounds interesting! could you share some metrics related to the performance of your model / explained variance / adjusted R^2? Do you have sufficient airlines who have a tendency to drop prices instead of steadily increasing them the closer you get to the desired departure date?
And most importantly
Did you backtest your model and run a simulation before pursuing this project to test its viability?
Thanks for the thoughtful questions! Sadly our model is still in the process of being trained (utilizing GDS transaction data) and we can't provide you with any metrics that would validate the viability without being actively deceptive. However, we can say that the data we have seen, coupled with the fact that both Hopper and Google have existing flight price ML models, that there is something here.
As you can see from the link we've submitted, we have a landing page that is purposely built to help gauge user demand for our product. Getting sign ups is our core metric and focus to gauge user demand as we train our model in parallel.
Would love to stay in touch and keep you updated on our model's progress. Feel free to sign up for our mailing list at the posted link or PM me.
Hey all, my team and I are developing an app that uses machine learning to predict price drops in flight prices. We then use these predictions to sell tickets cheaper than the market rate. Happy to answer any and all questions and would love any feedback.
Good question. Gas will be a key input into any future model that we build, but given we are in the initial stages of building out our app, we are not baking in gas prices into our model.
