Ride Hailing Service Uber frequently charges us with surge pricing. And we can't know the what time we should hail a ride, to minimize the fares within a time frame. We wished to predict the surge prices at times we intend to hail a ride and use a 1 hour buffer time frame to plan our ride with minimal fares.
What it does
It uses previously housed data from Uber's estimates API collected over some period of time, to predict the surge at a particular time. It also tells you the current surge price. And it suggests you the time at which it predicts the minimal fare between 30 minutes and after 30 minutes of the time whose data has been requested. And it notify's you 10 minutes before the time when minimal price surge is expected
How we built it
Used the Uber's Surge API, with android app as the frontend and python Django server as the backend. The app sends the following data: time, day, start location, end location to the server, and the server uses Linear Ridge Regression on the housed data to predict the surge
Challenges we ran into
Deciding on which prediction model to use to predict. Not enough data was available to run our tests.
Accomplishments that we're proud of
We intelligently synthesized data based on daily Uber pricing trends and successfully ran our prediction model on it. Developed an android app within 20 hours.
What we learned
Hackathons are fun. Popular Prediction models. Importance of UI/UX. Android app development.
What's next for foobar
An API based on our project. Better prediction model, based on more independent variables. Better UI/UX for the app. More hackathons! \m/.