My open-source projects translate methodological research into tools for applied scientists, with a focus on handling missing data, analysing conjoint experiments, and generating synthetic data.
MIDAS: Multiple Imputation using Denoising Autoencoders
A deep learning method for accurate and efficient multiple imputation. MIDAS is available for both Python and R users.
cjbart: Heterogeneous Effects Analysis of Conjoint Experiments
cjbart is an R package for analyzing conjoint experiments using Bayesian Additive Regression Trees (BART), specifically focusing on inspecting heterogeneous treatment effects.
SyGNet: Synthetic data using Generative Adversarial Networks
SyGNet is a Python package for generating synthetic data for the social sciences using deep learning methods.