This summer, I worked with Professor Gad Allon to analyze the effects of the COVID-19 pandemic on the ride-sharing industry. It is clear that, like many other industries, ride-sharing companies experienced a significant shock to both their demand and supply as a result of the pandemic. The first part of our analysis consisted of determining which of these shocks was more significant: the shock to the demand of riders or the shock to the supply of drivers. We also analyzed whether the demand and supply shocks induced by the pandemic were dependent on the time of day. In order to help answer these questions, we analyzed the effect of the COVID-19 pandemic on the usage and magnitude of surge pricing. The surge pricing mechanism acts as an intervention that regulates the supply and demand within the ride-sharing industry when the demand for rides significantly exceeds the supply of drivers. Therefore, a decrease in its usage would indicate that the demand shock was relatively greater than the supply shock, and vice versa.

Using a dataset of 152 million trips that took place in Chicago, I created an algorithm that estimated the surge multiplier for a given trip using robust linear regression and Python Pandas. Then, I analyzed the change in the usage of surge pricing and the magnitude of surge multipliers before and after the COVID-19 pandemic using the Mann-Whitney U Test in each of the 77 community areas. I also analyzed whether the time-of-day distribution of surged trips changed as a result of the pandemic. In other words, I determined whether or not the demand and supply shocks were dependent on the time of day by analyzing the change in the proportion of surged trips that occurred during each time period within the day.

We found that in most community areas, the usage of surge pricing decreased as a result of the COVID-19 pandemic. In other words, the demand shock was more significant than the supply shock, suggesting that the ride-sharing industry is more susceptible to demand shocks than supply shocks. We also found that the time-of-day distribution of surged trips changed because of the pandemic. Specifically, there was an increase in the proportion of surged trips that occurred during AM Off Peak hours (12am to 7am) and a decrease in the proportion of surged trips that occurred during PM Peak hours (5pm to 8pm) and PM Off Peak hours (8pm to 12am). This could have been caused by a combination of two reasons: (1) drivers have begun to work later hours as a result of the pandemic and/or (2) the distribution of the demand for rides has shifted earlier in the day.

Through this research experience, I learned a lot more about data analysis, specifically using Python Pandas and Statsmodels. Much of the work that I did consisted of analyzing the structure of the data and creating algorithms to either transform the data into different forms or make predictions using regression models. Because I did not have any prior research experience, I learned a significant amount about how to make hypotheses based on visual analyses of the data and support those hypotheses with more in-depth statistical analyses. Overall, this summer was a great introduction into how research is performed in the business world.