CART

A model that predicts the total number of border crossing based on Year and Measure of Transport. In other words, when given Measure of Transportation and Year, the intention is that this model could potentially rightly predict, the total number of border crossing in that year using that means of transport. The model is based on CART algorithm.

For the purpose of simplicity and to make our CART graph look not necessarily huge, changes were made to the values of Transportation, which is one of the explanatory variable in CART model. There were different values for Transportation that conveyed the same information more or less. related values for Transportation were grouped into one boarder value that denotes the same thing overall. In particular, there were 12 unique values that Transportation could take, namely, Bus Passengers, Buses, Pedestrians, Personal Vehicle Passengers, Personal Vehicles, Rail Containers Empty, Rail Containers Full, Train Passengers, Trains, Truck Containers Full, Truck Container Empty, Trucks. All related values were changed so that they are represented by one single value. For example, Trains, Truck Containers Full, Truck Container Empty, Trucks are now represented by just Truck.

In the CART tree below, following are the meaning of abbreviations used.

Bus: Bus

Pds: Pedestrians

Ral: Rail

Trn: Train

Trc: Truck

Each leaf node denotes one subclass that satisfies the conditions defined by its predecessor nodes in the path leading to it from the root. For example, the model predicts that after 2007, the number border crossing done by the means of Personal Vehicle is 141000000 while before 2007 is 90000000. As such, one application of this model would be to predict future border crossings in the United States.

As of now, this model doesn't actually capture everything that goes into what causes the border crossings.