Why Carpool?

Inspiration

In a busy city like NYC, countless Taxi trips happen everyday. Imagine two people travel at the same time from the same place to the same destination, they can actually ride only one taxi insdead of two.

However, they don't know each other... They ride two taxis seperately.. And they make the roads more congested by adding a taxi on the network...

Carpool or Not Carpool?

But is it necessary for travelers/comuters in NYC to carpool? Does Carpooling help improve the traffic condition? The following is person trips happened in the first week of June 2015.

How to Carpool?

Where to Start with?

We analized the June 2016 NYC Taxi Trip Records and find out:

Distribution of Taxi Trips Count Taxi Trip Frequency by Location

and we decided to focus on these Hot Areas, where there must be numerous taxi trips happening every day.

Hot Areas

So we choose this region for our test: region range (40.74,-74) cross (40.77,-73.96), time range: A busy morning on May 1 2015

How to Maximize the Total number of Carpooled trips

We have created an Integer Programming Model

How to Solve the Model

We have created an NYC Taxi Trip Carpooling Assitant using Matlab, which can be downloaded from here.

The source codes are published on Github

We used May 1 2016 records as the input data for this App. And the results are shown below:

Paired Trips

Paired Trips

Practical Meaning

Whats Next?

One extension of the application is to add weights to the edges, for example if two people are often paired together they are a strong pair. Another extension is giving regular commuters priority to be paired.