Reinventing the travel app

Photo credit Unsplash / Annie Spratt

I am currently a student at Lambda School and to graduate, we have to work on a team project that involves students from all tracks. As a data scientist, it was illuminating being able to work with iOS engineers, web devs, and UX designers. In our work, think of each of our parts like a puzzle piece. It is amazing seeing it all come together to create something so great.

Most travelers have a problem, it can be overwhelming planning a trip. When traveling, you have one site for accommodations, another app for events to attend, and even another app for flights or car rentals. To make this easier, we created an app called Resfeber. It is an app that takes all these things and puts them in one manageable place. Your hotel confirmations? Check. The list of awesome sights you want to see? Check. How much will it cost you in gas to drive all the way across the US? Check. It makes the traveler’s life easier so you can relax and enjoy planning that trip. When we first started finding the data to use for the app, I started to get kind of nervous. It seemed as though there are some components that do have limited data. As data scientists, it may seem like we are wizards, but alas, we are not. However, we are very resourceful and were able to fill in the gaps where needed.

For the data science team, we broke down our product into a couple of different pieces to work on. These included sourcing the data, cleaning the data, and then creating models.

How we broke down our gas prices component which is a huge part of the app

The technical mumbo jumbo

The main thing I worked on was gas prices. I sourced the data, cleaned and processed it, and then created a model that could predict gas prices.

A linear model for gas predictions

We wanted to be able to estimate the cost of gas for a trip to help people with their budget. One challenge though was finding the data to use. The data for gas prices is so limited, and the best dataset I found only included prices as a national average. What I started thinking though that since this was the average, that at least gave us a rough estimate instead of trying to calculate exact prices. A challenge we faced as a team was the same thing, limited data. The other data scientists that worked on Air B&B price models ran into the same issues. There just was not enough data to create prediction models for the whole US, so we are starting out using this model in select cities where the data can support it.

A beautiful product

We are currently finished with the data science portion of the app. Here is a list of the two features we have created:

  • Gas price prediction model that can predict the cost of gas for that trip.
  • Air B&B prediction model that can predict the cost of the accommodations for your stay.

In the future, I would love to see a feature with weather predictions, although that would be extremely hard since the weather is not truly that predictable. There are patterns that can happen but we cannot predict the day to day weather.

During the duration of this project, I have learned how to be a better teammate and data scientist. I have learned that it is okay to try new things and not to be afraid of failure. If something does not work, do not give up. Move on and try something new. Beautiful things take time to create.

I think this project helps further my career goals from an analytical point. There was a lot of analysis that went into the different parts of our projects and I love it. Getting down to the nitty-gritty and looking at the data to figure the best way to use it was fun.

--

--

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store