σ StatsDoge Causal inference workflows
59peers Log in Sign up
8
σ Building block · used in 1 workflow

Quantile forest

added by @grf · Jun 2, 2026
OTHER Machine LearningRandom Forest
Source grf — Athey, Tibshirani & Wager
Summary by StatsDoge

Conditional quantile estimation — the full predictive distribution, not just the mean.

You're looking at a building block — one of the estimators a workflow uses inside its pipeline. You reached it from a workflow step; it's used in 1 workflow (listed below).
⌥
Open repo ↗
Paper ↗

⚠️ Unofficial community write-up of a method from grf-labs/grf (pinned at v2.6.1). Not affiliated with the grf-labs authors — this summarizes the public documentation for demonstration. All credit & copyright belong to the original authors (Athey, Tibshirani, Wager, et al.).

What it does

Estimates conditional quantiles Q_τ(Y | X) for any set of quantile levels, using a quantile splitting rule (splits that separate the outcome distribution, not just its mean).

qf <- quantile_forest(X, Y, quantiles = c(0.1, 0.5, 0.9))
predict(qf, X.test)

Use it for

Prediction intervals, heteroskedastic outcomes, and detecting where the spread (not the level) of Y changes with X.

Used in these workflows (1)

Discussion (1)

  • 6
    @forest_fox · Jun 2, 2026

    Great for prediction intervals when the noise is heteroskedastic. The mean forest hides exactly the structure you care about.