Despite the fact that there are a greater number of airlines than most people can count, very few seem to differentiate themselves in regards to pre-purchase interaction with the customer. This results in the rise of middle-man sales dealers (expedia, travelocity, etc) willing to fill this void.

We believed that by designing a search methodology that is built from the ground up, predicated on how consumers approach planning a vacation, we could yield a win-win scenario; first, being able to enlighten travellers to the many options they have based on their personalized constraints (schedule, price) and preferences (general geographic location, type of vacation, specific city preference), and second, to allow airlines to better market their reduced fare, last minute getaway packages in an attempt to minimize the cost of empty seats on flights.

What it does

On the topic of impacts, North Star does two things. It helps consumers plan travel in a more intuitive and natural way than ever before, and it helps prevent unnecessary losses for airlines.

Airlines participate by allowing us access to their data or APIs (JetBlue and priceline supplied us with data sets here), and consumers benefit by using our site.

*TL;DR: People can take more vacations. Airlines lose less money. *

How we built it

So many of these things were done concurrently, but if we were to break them up into individual actions, they'd be as follows.

  1. Downloaded data sets from JetBlue and Priceline 1b. Parse wikipedia for accessory data (Flight code --> Airport Name --> Airport City --> City Longitude/Latitude)
  2. Set up server on Azure
  3. Populate SQL server
  4. Integrate facebook
  5. Write search algorithm / build front end for user inputs 5b. Design all static images (radio button icons for flight properties, company logo, etc).
  6. Designed map interface, planned how the data would influence dynamic changes in the map 6b. Used beaker and Priceline data to show analytics on charts so that consumers can have a better idea of how the current flight prices compare to the historical values, or to compare prices by airline.
  7. Built the 'infinite scroll' functionality and polished up front end

Challenges we ran into

Not having an API to dynamically call information makes for a lot of wasted time when we consistently need to remake/repopulate the database.

We messed up with how we imported the dates and times of flights in our SQL database. This was huge, since times and dates are 2 out of the 3 most crucial pieces of information (after location).

Python really really really doesn't like unicode. Our parser broke 2-3 times when it encountered hyphenated airport names or accented letters.

Accomplishments that we're proud of

1. Scalability *(Database layout, algorithm efficiency) This is our team's third consecutive data hack submission at a hackathon, so concepts like efficiency and resource utilization are things that are prioritized whenever possible. We believe that after multiple projects that utilized 'bandaid' fixes under the 'Fuck it, ship it' mentality, we've finally started to do things the 'right' way, and in a way that is scalable. *2. Innovation We were able to attack a seemingly stale methodology from a fresh perspective. Disruption is cool, and being the new kid on the block is always fun. *3. We learned a lot. * Beaker, Facebook API, Microsoft Azure, and tons of SQL troubleshooting. Despite the fact that we all have very distinct and unique skills with a bit of overlap, we all ended up doing far more work that was outside of our respective comfort zones than ever before for this multi-leveled hack.

this memorable quote summarizes it pretty well: Brian Chen: "Please, no more SQL. I can't handle any more." Nick: "You gon' learn today, you gon' learn real good about SQL, inside and out".

What we learned

Take care of your most important data points/structures first. Data is love, data is life - except when it takes a combined 6 hours to repeatedly repopulate your database after improper planning. ALWAYS think your database through ahead of time!

Beaker is awesome. Nick loves working with data, and for him, Beaker is a gift from the heavens. He wishes it could work with MATlab too, but he doesn't want to push his luck.

What's next for North Star

1. Increase the amount of data that we have! We were given finite data sets by sponsors, so by having live access to either APIs or more time to parse websites, the data would only further serve to increase the effectiveness of North Star.

*2. Continued Brand Differentiation * Despite the fact that many other travel sites supply tons of functionality with their seemingly endless filters, it all comes across in a very cold and emotionless way. For most people, travelling is a fun (or at least emotional) experience, and we believe that the process leading up to it should be no different. If North Star were to be continued, it should embrace its different style of approach by getting user feedback and continuing to make things 'shiny' and 'more polished'. Although it's not frequently spoken about, user experience on a travel site could be the difference between a consumer packing their bags for an overnight getaway or sitting home on their couch to Netflix and chill.

3. Innovative Data Processing/Empowerment By incorporating Priceline data (which contains data from a multitude of airlines) and combining it with Beaker, we have the power to offer a platform with a stronger price analysis tool than any other competitor.

4. Integrate Consumer Behavioral Science Beyond conscious preferences of the consumer, a multitude of market research papers reveal trends that differentiate spending patterns within the airline industry. By having access to more data from our users (geographic location, approximate income level by zip code, race, age), we could use these demographic values to alter the meta-rankings generated by our database search algorithm, showing results that are more in line with what the consumer ultimately will be interested in considering. One example of this, is that millenials are 60% more likely to book a flight that is discounted than nonmillenials. Another, comparing the same two groups, is that millenials are significantly more likely to pay a premium to upgrade to an airline seat with more leg room, while also disproportionately prioritizing budget-oriented airlines. We successfully integrated the Facebook API to easily get demographic information about our consumer, however, without a significant amount of time (or money) to access these market research studies, we didn't have much to work with during the time of the hackathon to implement justified meta rankings. We were mindful of this though, and set up our database in such a way that we could easily implement these factors as multipliers/scalars if continued.

Share this project: