Since the 2016 election, inland U.S. Border security has been the huge topic. The construction for the new border wall has started and the tension between Mexico and U.S. has intensified along with it. Many people predicted not only the decrease in number of illegal border entry but also the decrease in number of legal border entry which could hurt the tourism and discourage the trade across the borders. Currently in end of 2019, 3 years from 2016 election campaign, what can we learn from the statistics of inland U.S. Border entry? Did the political agenda affect the way people come into our country? I would like to answer these questions through visualizing the findings.
I used the dataset from Kaggle.com . This dataset is originally collected by U.S. Customs and Border Protection (CBP) every quarter which then gets cleaned, assessed and maintained by the Bureau of Transportation Statistics (BTS). It contains the statistics for inbound crossings at the U.S.-Canada and the U.S.-Mexico border at the port level for every month from beginning of 1996.
The original dataset includes 349,000 rows with 7 Columns as following: Port Name, State, Port Code, Border, Date, Measure, Value and Location. Measure is a method of transportation used for border entry and it has 12 different categories: Bus Passengers, Buses, Pedestrians, Personal Vehicle Passengers, Personal Vehicles, Rail Containers Empty, Rail Containers Full, Train Passengers, Trains, Truck Containers Empty, Truck Containers Full, Trucks. Values column includes the total number of crossing.
It is important to be aware that this dataset doesn’t count the number of unique vehicles, passengers or pedestrians but rather count the number of crossings. For example, same truck can go back and forth the border many times a day and data for each time will be collected. Also this data doesn’t include the nationality of the passengers or pedestrians nor the reason for the border crossing.
I used R package dplyr to clean the data. Criteria for Border Name was shortened as Canada and Mexico and Location column was divided into two sections: Longitude and Latitude. Also year 2019 was excluded from the analysis as the data is not complete yet and that will not provide a good insight for this project. The exploratory data analysis (EDA) was done mostly using R package ggplot2 and leaflet. I used the ShinyDashboard to show the visualization and shinyapps.io as a server to present the findings.
ShinyApp / Analysis
First I wanted to observe the location of border ports and their distribution across the U.S. to see the big picture.
There are total of 116 ports used in this dataset. Among those, 89 ports are in U.S.-Canada Border and 27 ports are in U.S.-Mexico Border. So there are about 3 times more ports in U.S.-Canada Border compared to U.S.-Mexico Border.
However the number of total incoming from 1996 to 2018 showed total opposite as there were about 7 billions of total border crossing at U.S.-Mexico Border while there were about 2.6 billions of total border crossing at U.S-Canada Border. Even though there were more ports available in the northern border of U.S., there were less people coming in.
Now moving on, I wanted to find out the methods of border entry and how it looks different between U.S.-Canada border and U.S.-Mexico border.
Overall, the most of the border entry method was by using personal vehicles. Here, the Personal Vehicles count the numbers of personal cars entering the border whereas the Personal Vehicle Passengers count the number of people that were in the Personal Vehicles. Next high value was surprisingly the Pedestrians. Bus Passengers and Trucks came after.
When the measure was compared between two borders, I could see a difference where the Mexico Border has a significant number of pedestrians whereas the Canada Border does not.
When I looked into the number of entry in different states, I could find the similar trend from the number of entry by two different Borders. There is more entry from the southern U.S. border especially in Texas, California and Arizona. From the northern U.S. border, the large number of border entry was from New York and Michigan.
Most of the southern states had a similar trend of entry transportation as Texas as it is shown on the left graph above: mostly containing the Personal Vehicle and Pedestrians. For the northern states, it looked similar to the New York as it is shown on the right graph above. This indicates that the northern and southern borders differ a lot in terms of number of people walking into our country.
Two exceptions to the statement were Alaska and Ohio. Alaska had more significant number of Bus Passengers and Train Passengers suggesting that most of the people coming into the U.S. from Alaska border ports are travelers. Ohio, the state with the lowest number of border entry, was reported with only one method of transportation: Personal Vehicles.
When I looked into the change in number of truck entering the U.S. in U.S.-Canada and U.S.-Mexico Borders, I saw some patterns. First, both Canada and Mexico sides had a sudden drop of number in certain year such as 2009. That was the year of global financial crisis. Trucks can be used for in-land trades and it is obvious from the graph above that the economy has a big impact on the border entry. Second, I was able to see the increase in number of trucks entering at the Mexico border, possibly suggesting a better trade condition between U.S. and Mexico.
As I wanted to see the impact of the 2016 election and the increase in border security issue, I looked at the number of Pedestrians and Personal Vehicle Passengers over the years. Surprisingly, unlike what I have guessed that issue of border security and building the border wall would discourage the number of legal border entry in U.S.-Mexico border, the statistics show the increase in border entry.
The number of incoming buses and bus passengers into the U.S. seem to be decreasing in both Canada and Mexico sides. Usually buses can be used for the tourism and as other methods of traveling such as flight and train have advanced over the years, use of bus as a method for traveling seems to have declined.
Two methods of transportation used for border entry that look distinctly different were Pedestrian and Train Passengers. As seen in previous graphs, most of the pedestrians are coming into the U.S. using U.S.-Mexico border as it is more accessible to walk across the border in the southern side of the States. Train is used more often in U.S.-Canada border and the number of its use has been increasing. This increase in use of train can be related to the decrease use of bus as a traveling method.
Different methods of transportation into the U.S. also show unique trend when the data was looked according to the months. Transportation methods that can be used for trade such as Truck, Truck Container Full, Rail Container Full seem to have a steady number of incoming in both Mexico and Canada borers no matter what month it is. This indicates that the business related border entry stays steady in all-year-around.
However this changes, when it comes to the transportation related to traveling such as Personal Vehicle Passengers, Bus Passengers and Train Passengers. Number of border entry in U.S.-Canada increases significantly in summer months while number of border entry in U.S.-Mexico stays around same all-year-around. As the northern border get a harsh cold winter, it is obvious to have more travelers in summer months.
Conclusion / Further Research
From the visualization and statistics, the number of border entry depends on the economy and business rather than the politics. However this dataset itself can’t explain the reason behinds the change in number of border entry as there are still many factors that need to be considered. For the further research, I would like to obtain the data regarding the citizenship of people that enter the U.S. borders as well as their intension or the reason for the entry. This can furthermore support how economy or tourism change the trend in the border entry.
Thank you for reading my findings in U.S. Border Crossing Entry data. If you are interested in looking at the dataset I used, my ShinyApps, and the code, you can follow with the links below.