The Bayesian Bandit -

Sold! Building a prototype Graph Neural Network for Ad Relevance

Ad revenue has become the primary lifeline for many modern companies. To optimize this stream, platforms don’t just show ads at random; they enlist hundreds of engineers and…

Eat More SDID? Testing Chick-fil-A Mobile Thru using Synthetic Difference-in-Difference

causal inference

Air PSM: Measuring the effects of dynamic pricing on host revenue

causal inference

A core component of vacation planning is reserving accommodations for the duration of a stay. Many vacationers use platforms like Airbnb to immerse themselves in the local…

DJ RD: Using regression discontinuity to measure music personalization

causal inference

A/B testing is the usual standard for understanding the effects of different treatments. A prominent use case is testing potential new product features. While A/B testing is…

(AI)rline Pricing: Estimating Willingness To Pay

causal inference

Dynamic pricing has been the dominant approach in airline revenue management systems for decades. This method has continually improved with advances in new algorithms and…

There and Back Again: Optimizing Home-Swapping on Kindred

In this post, we introduce Kindred, a new way to travel [1]. We first espouse the company mission, vision, and values through the lens of business strategy. We continue this discussion by moving into the more technical details of…

The Impact of Sports on University Philanthropy: A Causal Analysis

causal inference

In this post, we present a scenario where a university seeks to identify the causal effects of sports passes on alumni donation likelihood. Using a synthetic dataset…

The Doomscrolling Algorithm: Optimizing Content to Maximize Revenue

reinforcement learning

In this post, we present a prototype recommendation system based on Deep Neural Networks for YouTube Recommendations [1] and Explore, Exploit, Explain…

Optimizing Optimization: A Tutorial on Dispatch Algorithms with Applications in the Gig Economy

In this post, we present an implementation for Next-Generation Optimization for Dasher Dispatch at DoorDash [1] and Us…

Do Scanner Guns Decrease Checkout Time? An Analysis Using Diff-in-Diff and SCM

causal inference

In this post, we present the situation of a grocery store chain looking to decrease checkout-time by supplying scanner guns at self-checkout registers. We demonstrate first…

Multiple Treatments in Marketing Messaging with Non-Compliance

causal inference

In this post, we provide an implementation for the paper Uplift Modeling for Multiple Treatments with Cost Optimization [1]. We provide this implementation in the context of a marketing department testing which marketing campaign is most effective for a given loyalty tier. We continue with this…

Agentic Marketing: Optimizing Out-of-app Messaging using Causal Inference

causal inference

reinforcement learning

In this paper, I demonstrate a prototype for utilizing reinforcement learning in out-of-app (OOA) messaging optimization to…

An implementation of the ITCC algorithm in Python

Clustering is a useful tool in data mining whereby data points are collected together based on some predefined similarity metric. This results in a more digestible data…

An analysis of Uber fares

Uber is the current leader in the rideshare market. With millions of people using the platform everyday, the company must prioritize resources to understanding their…

Forecasting Enrollment Size by Class at TAMU

Forecasting enrollment for individual classes is important for departments and colleges at universities. It allows them to properly prepare adequate resources for students…

An Analysis of TA-MU Hotel Booking Data

TA-MU Hotels is a small, privately owned hotel business. Currently, TA-MU Hotels owns two hotels: One in the heart of Manhattan, NYC and another in Cancun, Mexico. TA-MU…

Welcome to my blog!

This blog focuses on the use of statistics, particularly bayesian statistics, in the application of business problems. In this post…