Author: gfaletto
-
PRESTO accepted to ICML 2023
I’m excited to announce that “Predicting Rare Events by Shrinking Towards Proportional Odds” has been accepted to the Fortieth International Conference on Machine Learning (ICML 2023)! In the paper, we propose PRESTO, a novel method for improving classification in the class imbalance setting. You can read my brief summary of the paper on Twitter.
-
cssr R Package
In a 2022 research paper that I wrote with my advisor Jacob Bien, we proposed a novel feature selection method called cluster stability selection. Cluster stability selection is a method for identifying features that are useful for predicting a response variable. It has applications in medical research (including genomics and genetics), economics, analyzing survey data,…
-
My Math Review Notes
I recently took the first-year screening exam for Ph.D. students in the statistics group in the Department of Data Sciences and Operations at Marshall. Since I started applying to grad school, I’ve been writing up review notes to help me with math. Originally the purpose of the notes was to help me study for the…
-
Our Entry in the OCRUG Hackathon 2019
I was a part of Team Save the WoRld along with Faizan Haque, Javier Orraca, Sam Park, and Shruhi Desai in the OCRUG Hackathon 2019 held at UC Irvine on May 18th and 19th. (In fact, I am writing this blog post at the tail end of our time before we present our results!) The…
-
Presentation on Multi-Task Learning
Today I gave an in-class presentation at USC on two papers in multi-task learning (or multivariate regression–linear regression when the response is a vector rather than one number). You could simply train a separate model for each response, but when the responses are related, there are advantages to considering them all at the same time…
-
The McCarthy/Fader/Hardie Model for Customer Retention
In another post, I described how I fit a model to predict how well Buffer—a digital subscription-based firm that publicly releases much of its financial data—retains its customers. I used a methodology developed by Daniel McCarthy, Peter Fader, and Bruce Hardie (paper available for free download here) to fit the model. In this post, I’ll fill in…
-
Which Customers Stick Around? Predicting Customer Retention
One of the most critical priorities for any business is retaining their customers for as long as possible. You want to keep your customers happy so they come back to spend more. Data science can help businesses keep customers longer. Using business data, a data scientist can develop a model that predicts if and when…