Cailin & Elise: Amazon Robotics
This summer we were selected to be interns at Amazon Robotics headquarters in North Reading, MA. Amazon Robotics. Due to COVID-19, our internship became virtual and we were assigned an independent summer project. Despite the challenges of working remotely and during quarantine, we had a great experience and learned a lot about Python - a new computer language for us - and managing data in a professional manner.
The project objective was to perform an analysis of Amazon’s competitors in the brick-and-mortar retail landscape and investigate their store locations relative to population density. To accomplish this, we were asked to create a computer program using PYTHON that combined US Census and Amazon data sources. We met once or twice a week with Tye Brady (Chief Technologist at Amazon) to discuss our progress and get his input and advice on the results and how to move forward.
In our kick off meeting with Mr. Brady, we were given an Excel spread sheet with information about competitor’s retail stores and their placement throughout the United States (U.S.). We also downloaded information directly from the 2019 U.S. census. With this information, we performed an analysis of the data and developed conclusions (a few statistics). Going into it, we spent a few weeks learning the coding language of Python. After that we collected our data together to compress it and make it as efficient as possible. Our final product (still in progress) is a report explaining our work and findings.
Since there were two of us collaborating on the project, each week we would decide on the parts to split up or work together on. Our first goal was to see how many U.S. counties out of the total 3,141 contained a competitor’s store. We created a list and deleted duplicates to get the end goal of the number of stores. Then, we linked county population data to the counties that contained a store. We used this information to sum the total number of people living in a county that contained the store. We then took this number and divided it by the total U.S. population in 2019, to get a percentage showing the amount of people in a county with a store. Our final task was to find the average distance from the most populated town in a county to a store and create a national average. Using census information from all the towns and counties in the U.S., we were able to generate a function that mapped out the difference in coordinate points over the data frame to get a distance in miles. After doing this for every store, we then created an average for all stores.
In the end, we were able to combine the store data with the county data and were able to sum the population of the counties that have a store. We combined all the data to eventually create a master data frame with all the basic information we used. To create a visual for our project we used Google Maps to plot the points of every competitor store in the U.S. It allowed us to create an interactive map that can show where the stores are located. Our final goal was to give Tye Brady a report of our work this summer to share with his colleagues.
We want to thank Mr. Quinlivan at Amazon Robotics and Mr. Schenker for figuring out a way to do a summer project when our internship was no longer possible. We also want to especially thank Mr. Tye Brady for mentoring us through the project. Mr. Brady taught us invaluable lessons about coding and analysis and working on Zoom. He has inspired us to further pursue career opportunities in computer science and hopefully to work with Amazon Robotics – in person - in the future.