Delay Predictor

Machine learning based flight delay predictor

Flight Delay Predictor from Upside Business Travel is a machine learning based product that attempts to predict the likelihood your flight is to be delayed. The algorithm is trained on historical flight delay information from the FAA and factors in both historical and forecasted weather and the current state of the National Airspace System.

Discussion
Would you recommend this product?
10 Reviews5.0/5
Hey @ian_cassidy, There seem to be a lot of apps for getting money from the airlines if your flight is delayed. Is this something you've thought of integrating into this project?
Hi @jacqvon, we've seen a lot of these other apps, but what we are trying to do here is help our customers make better decisions so that they can get to their destination on time. If you book with Upside, our customer service team is immediately alerted if your flight becomes delayed and will offer to help put you on an alternate flight. We are working towards integrating the prediction engine into this process when we have high confidence that your flight may become delayed.
Upvote (2)Share
This is super interesting. Here's an example of a flight from SFO to LAX Wednesday afternoon: Although I'm not sure I'm willing to risk arriving late based on these predictions. 😅
Hey @rrhoover, thanks for trying out the Delay Predictor! It looks like the model is pretty uncertain about this flight since it is ~2 days out, which makes sense. Keep monitoring the probabilities as it gets closer to the departure time to see how things are changing!
Upvote (1)Share
Hi @ian_cassidy could you go into a little more detail on how the prediction algorithm works? I'm a student and am super curious about how you guys built this!!
Hi @malavika_vivek, we actually have a blog post about the algorithm and I just added a link above. Here it is: https://engineering.upside.com/a...
Upvote (1)Share
Hey ! Very interesting. I know that google is working on something similar (https://www.blog.google/products...) Can you tell me what are the weaknesses of their prediction engine compared to yours ? Thank you !