ocf: Ordered Correlation Forest

In ocf: Ordered Correlation Forest

Ordered Correlation Forest

Description

Nonparametric estimator for ordered non-numeric outcomes. The estimator modifies a standard random forest splitting criterion to build a collection of forests, each estimating the conditional probability of a single class.

Usage

ocf(
  Y = NULL,
  X = NULL,
  honesty = FALSE,
  honesty.fraction = 0.5,
  inference = FALSE,
  alpha = 0.2,
  n.trees = 2000,
  mtry = ceiling(sqrt(ncol(X))),
  min.node.size = 5,
  max.depth = 0,
  replace = FALSE,
  sample.fraction = ifelse(replace, 1, 0.5),
  n.threads = 1
)

Arguments

Y	Outcome vector.
X	Covariate matrix (no intercept).
honesty	Whether to grow honest forests.
honesty.fraction	Fraction of honest sample. Ignored if honesty = FALSE.
inference	Whether to extract weights and compute standard errors. The weights extraction considerably slows down the routine. honesty = TRUE is required for valid inference.
alpha	Controls the balance of each split. Each split leaves at least a fraction alpha of observations in the parent node on each side of the split.
n.trees	Number of trees.
mtry	Number of covariates to possibly split at in each node. Default is the square root of the number of covariates.
min.node.size	Minimal node size.
max.depth	Maximal tree depth. A value of 0 corresponds to unlimited depth, 1 to "stumps" (one split per tree).
replace	If TRUE, grow trees on bootstrap subsamples. Otherwise, trees are grown on random subsamples drawn without replacement.
sample.fraction	Fraction of observations to sample.
n.threads	Number of threads. Zero corresponds to the number of CPUs available.

Value

Object of class ocf.

Author(s)

Riccardo Di Francesco

References

• Di Francesco, R. (2025). Ordered Correlation Forest. Econometric Reviews, 1–17. \Sexpr[results=rd]{tools:::Rd_expr_doi("10.1080/07474938.2024.2429596")}.

See Also

marginal_effects

Examples

## Generate synthetic data.
set.seed(1986)

data <- generate_ordered_data(100)
sample <- data$sample
Y <- sample$Y
X <- sample[, -1]

## Training-test split.
train_idx <- sample(seq_len(length(Y)), floor(length(Y) * 0.5))

Y_tr <- Y[train_idx]
X_tr <- X[train_idx, ]

Y_test <- Y[-train_idx]
X_test <- X[-train_idx, ]

## Fit ocf on training sample.
forests <- ocf(Y_tr, X_tr)

## We have compatibility with generic S3-methods.
print(forests)
summary(forests)
predictions <- predict(forests, X_test)
head(predictions$probabilities)
table(Y_test, predictions$classification)

## Compute standard errors. This requires honest forests.
honest_forests <- ocf(Y_tr, X_tr, honesty = TRUE, inference = TRUE)
head(honest_forests$predictions$standard.errors)

## Marginal effects.
me <- marginal_effects(forests, eval = "atmean")
print(me)
print(me, latex = TRUE)
plot(me)

## Compute standard errors. This requires honest forests.
honest_me <- marginal_effects(honest_forests, eval = "atmean", inference = TRUE)
print(honest_me, latex = TRUE)
plot(honest_me)

ocf documentation built on April 4, 2025, 4:44 a.m.