# Goizueta Student Interview: Libby Ying BBA18, Winner of the 4th Annual Predictive Analytics Competition

Libby Ying, BBA18, is this year’s winner of Goizueta’s Annual Predictive Analytics Competition, organized by Assistant Professor Nikolay Osadchiy (Information Systems & Operations Management). Participants were challenged to analyze data on the automotive industry and produce a forecast for vehicle sales. Atlanta-based Automotive Ventures LLC (founded by Goizueta alumnus Steve Greenfield) and alumnus Vu Pham sponsored the competition.

GBL: You entered the Predictive Analytics Competition as a one-person “lone wolf” team – what inspired you to join the competition? Do you have a lot of experience with predictive analytics?

I really enjoy information systems. I was inspired to join the competition because forecasting a large amount of data such as BMW data intrigued me. Additionally, my family and I love watching baseball. There is a lot of data and statistics out there on every aspect of baseball. For a competition, I once used baseball data/statistics to try to predict the outcome of baseball games and who will win the World Series. Thus, I love predictive analytics.

GBL: According to the judges, your model “had the lowest amount of variance” and “was the best fit for the data.” What led you to consider using a “top down” approach in your solution?

I did try to use multiple linear regressions but there was not a lot of time and I was working alone. Thus, after spending more than three hours and getting nowhere with multiple regressions, I decided to take a deep breath and try to look at the problem from a different perspective. In multiple regressions, you are trying to predict an unknown (when a car will be sold) based on certain known factors, such as model, time in the lot, where is the car now, etc. The key part is to find the most important factors then you can predict with high precision the selling date of each car. This is what we call a “bottom-up” approach that is predicting one car at a time. Instead, I decided to look at some high level statistics. We have 4 months of sales data. After creating a Pivot table with all the data and doing some analysis, I noticed that there is a trend in the distribution of time to sell a car. For example, most cars were sold after two months after arriving at the dealership. After some more analysis, I found out that I could predict the percentage of cars that will be sold in January based on the number of cars that arrived in the dealership from September, October, November or December.

GBL: What was your research process for solving this problem? What resources were helpful to you? Did you use any library resources or consult with a business librarian?

Because I was only given two days to analyze the data, I had to think quickly and work through the night. While at the Goizueta Business Library, I researched different possible ways to solve this problem using the business library databases. A lot of sources told me to look at probability. I then looked at some old statistics books at the library to get some inspiration.  While sitting in the library, I came up with this “top-down” approach while looking at the probability section of the statistics books. Business Librarian Susan Klopper helped me greatly with research for my project and with life in general.

GBL: Your great performance in the competition led to an offer for a paid internship with Automotive Ventures. Congratulations! What are you hoping to gain from the experience? Did you ever imagine you would work in the automotive industry? How do you see this experience impacting your career future?

Throughout my career, I don’t want to just have one concentration. I want to have experience in Information Systems, Consulting, Finance and more and use all those skills in various jobs throughout my career. Thus, I am very excited to intern at Automotive Ventures. I hope to gain more experience with forecasting. I did not imagine myself working in the Automotive industry but I am extremely happy, excited and grateful for the opportunity.