Blog

List of blog posts

Uncertainty-aware model comparison

Comparing model performance rigorously and meaningfully using statistics.

machine learning

deep learning

statistics

meta-science

Oui-Love Plots: Outcome-informed Love plots for covariate balance in causal inference

Augmenting the good old Love plot by incorporating covariate-outcome importance measures.

causal inference

visualization

Sequential trial design for causal inference

Aligning time-zero to reduce time-related biases by taking simplest solution possible and awkwardly repeating it until it’s great.

causal inference

study design

Hello Quarto!

I made a personal website. With blogs and stuff.

Causal inference is a mindset

Causal inference from observational data is a mindset, not a set of tools.

causal inference

Causal Inference with Continuous Treatments

Generalizing inverse probability weights for non-categorical treatments.

causal inference

Why we care for covariate balancing in comparative studies

Balancing variables in statistical comparative analysis is a proxy, not a goal.

causal inference

A visual way to think of macro and micro averages in classification metrics

Explaining what macro-average and micro-average metrics are.

machine learning

visualization

Using machine learning metrics to evaluate causal inference models

Reinterpreting known machine learning evaluations from a causal inference perspective, focusing on ROC curves for propensity models.

causal inference

machine learning

The case against Agile research

Ever popular iterative development approaches can sneak in unconscious-bias that can be harmful to the scientific process.

meta-science

Solving Simpson’s Paradox with Inverse Probability Weighting

A visual intuition on how the most popular method in causal-inference works, and how it solves one of the most popular paradoxes in statistics.

causal inference

statistics

Applying Deep Learning to Genetic Prediction

What classical methods for obtaining polygenic (risk) scores lack, and how deep learning might help mitigated these shortcomings.

genetics

deep learning