Hello Connected Universe,
Douglas here from Indianapolis, IN and recent grad from Notre Dame's Data Science master's program.
I first learned of Neo4j in the program and was pretty blown away. I previously had my own consultancy in the CRM space and struggled when having to implement asymmetrical relationships that are so effortless in Neo4j. Since then I have used Neo4j to model disruptive events on airline routes, and architected mobile app backends using Neo4j. My next tasks are to build ML models for the backend: recommender, fraud, language.
Nice meeting all of you,
Douglas
1 Like
Welcome to the community! Very cool projects!
Did you have a favorite algorithm for the airport route project? Did you add other factors like weather?
Yolande
Hi Yolande Poirier,
I've seen you recently in presentations! Hello again!
I very much leveraged the Betweenness Centrality algo on the project, weighted by number of flights on each route. And yes, we would add a weather node to the model. A weather node had attributes like lat/lng (the geographic center of the event) , type (snowstorm), category (quantifying severity based on National Weather Service guidelines), size (swag on its circumference). We then would calculate a probability of disruption based on this information and then insert that statistic back on the route relationship.
This was performed on a fixed snapshot in time, but operationalizing this idea we would in realtime update this model showing all in-progress flights and introduce disruptive events as they occur (say in a feed from the National Weather Service).
In closing, my model was much cooler than what Google did on their Big Data YouTube video. Ha, had to say that...thanks for checking in!
Douglas
Pretty cool. If you decide to make a video or blog about it, let us know. We’d love to share it.
