Research projects

The Amazon project aims to analyze the impact of customer ratings on the investment decisions of publicly traded companies. Data from Amazon.com from May 1996 to October 2018 was collected from the University of California San Diego (UCSD) data repository. This data consists of two parts, metadata and reviews, both of which can be linked through the product ID. The objective of the project is to link the products to the parent company that ultimately benefits from the sale. However, there are challenges that need to be addressed, such as the accuracy of brand names, missing values, availability of a comprehensive list of publicly traded firms over the years, reliable data regarding brands produced by the firms, and matching the brand names from the UCSD database with the brands produced by the firms.

The first step in the procedure was to come up with a list of public companies and brands by downloading public companies' names from Nanex and performing a Google search to find the related subsidiaries. The names of the subsidiaries were then compared with the public company name using the Fuzzywuzzy package in Python. The results of this process were then filtered to retain only relevant information. After an intense data manipulation, preliminary analysis shows that amazon reviews are (a) able to predict future returns, and (b) are correlated with the level of R&D and CAPX investment of companies that sell through amazon.