This is a NLP and machine learning case study predicting whether a project proposal would be accepted onto DonorsChoose.org, a crowdfunding platform for public high school teachers
Background: DonorsChoose.org is a nonprofit organization that connects donors with public school teachers to help fund classroom projects in the United States. Donors can choose specific classroom project or supplies to support, and can make donations of any amount.
Problem statement: DonorsChoose.org expects to receive close to 500,000 project proposals next year; therefore, there are three main problems that need solving:
- Scaling current manual process of screening 500,000 projects so that they can be posted as quickly and efficiently as possible
- Increase the consistency of project vetting across different volunteers to improve experience for teachers
- How to focus volunteer time on the applications that need the most help
Project aim: Predict the probability that a proposal will be accepted based on the material submitted by each applicant (i.e., text of the project description and its meta data).
Jupityr notebooks will go here after some cleanup, commenting, and reorganizing. The project was completed in April 2018 when I was exploring the data science space.