Metis Project 4
This project will focus on NLP and unsupervised learning techniques.
I will be working with Municipal Bond data collected from the EMMA website. General Obligation bonds are often not tied to an explicit project and may have multiple use of proceeds wrapped up in the offering document. The majority of the documents then are going to be very generic in language and use, often mainly focused on refinancing old issuences or basic government operations and repairs to buildings. However, there can be issues that do focus on more interesting topics such as affordable housing or economic development projects.
The goal of this project is to see if machine learning can help us to both find bonds that stand out from the perspective of impact investors and more specific details about what the use of proceeds will be used for.
Data_Collection_Final contains the code for collecting the offering documents in pdf from and saving them to a cleaner more usable pandas dataframe.
Project_Walkthrough_Final contains some EDA and the process of trying various topic modeling.