Capacity Predictor: A Machine Learning Approach to Ratings

Presenter: David Schemitsch, Columbia University, Adam Bradford, Columbia University

Year: 2020

Price: Members: $49, Non-members: $79

Description: Traditionally, capacity formulas have trouble conveying uncertainty in the estimate as well as incorporating conflicting wealth indicators into a coherent ratings framework. To resolve both of these issues, we use publicly available survey data from the Federal Reserve to create a machine learning model of capacity as a function of identified assets. This framework can be utilized by prospect researchers in the form of an app, or it can be used to screen your database. The presentation will cover the traditional way of rating prospects; thinking about ratings as probabilities;  an exploratory data analysis of the Federal Reserve's survey data; what went into creating our machine learning rating system; and a demonstration of the rating app.

Recent Stories
Defining and Evaluating the International Prospect Pool

Same-Page Strategy: Driving Results through Custom Profiles

How Organizations Can Leverage Wealth Screening and Data from Donors Acquired during the COVID-19 Crisis to Accelerate Fundraising Growth and Improve Retention