RLT Classification Tutorial

Overview

This page shows how to fit and predict a classification model with RLT.

Prerequisites

Install RLT, then load it with library(RLT).

Data Preparation

We generate continuous and categorical predictors with a binary outcome.

# (Optional) For reproducibility in this tutorial only.
# Detailed notes on random seeds are in the Seed & Reproducibility feature page.
set.seed(1)

# ---- Generate a small synthetic dataset ----
trainn <- 80
testn  <- 20
n <- trainn + testn
p <- 10

# Continuous + categorical predictors (last half as factors)
X1 <- matrix(rnorm(n * (p/2)), n, p/2)
X2 <- matrix(as.integer(runif(n * (p/2)) * 10), n, p/2)  # integers 0-9

X <- data.frame(X1, X2)
X[, (p/2 + 1):p] <- lapply(X[, (p/2 + 1):p], as.factor)

# Binary outcome via a simple logistic signal
logit <- function(x) exp(x) / (1 + exp(x))
prob  <- logit(-0.5 + 2 * X[, 1])  # simple signal using the first continuous feature
y     <- factor(rbinom(n, 1, prob = prob), levels = c(0, 1))

# Train / test split
trainX <- X[1:trainn, ]
trainY <- y[1:trainn]
testX  <- X[(trainn + 1):(trainn + testn), ]
testY  <- y[(trainn + 1):(trainn + testn)]

Fit a classification model

ntrees <- 200
ncores <- 1
nmin   <- 5
mtry   <- p/2
samplereplace <- TRUE
sampleprob    <- 0.80
rule    <- "best"
nsplit  <- ifelse(rule == "best", 0, 3)
importance <- TRUE

fit <- RLT(
  trainX, trainY, model = "classification",
  ntrees = ntrees, mtry = mtry, nmin = nmin,
  resample.prob = sampleprob, split.gen = rule,
  resample.replace = samplereplace,
  nsplit = nsplit, importance = importance,
  param.control = list(alpha = 0),
  ncores = ncores, verbose = FALSE
)

Predict

pred <- predict(fit, testX, ncores = ncores)

# Helper to extract a numeric or class vector
get_pred <- function(obj) {
  if (is.list(obj) && !is.null(obj$Prediction)) obj$Prediction else obj
}

# Predictions for train/test
train_pred_raw <- if (!is.null(fit$Prediction)) fit$Prediction else get_pred(predict(fit, trainX, ncores = ncores))
test_pred_raw  <- get_pred(pred)

# Coerce to class labels aligned with trainY levels
to_class <- function(x, ref_levels) {
  if (is.numeric(x)) {
    # Treat as probability for the "1" class; threshold at 0.5
    cls <- ifelse(x >= 0.5, as.character(ref_levels[2]), as.character(ref_levels[1]))
    factor(cls, levels = ref_levels)
  } else if (is.factor(x)) {
    factor(x, levels = ref_levels)
  } else {
    factor(as.character(x), levels = ref_levels)
  }
}

train_pred <- to_class(train_pred_raw, levels(trainY))
test_pred  <- to_class(test_pred_raw,  levels(trainY))

Evaluate

acc_train <- mean(train_pred == trainY)
acc_test  <- mean(test_pred  == testY)

# A compact summary
list(
  Train_Accuracy = round(acc_train, 4),
  Test_Accuracy  = round(acc_test, 4)
)
## $Train_Accuracy
## [1] 0.725
## 
## $Test_Accuracy
## [1] 0.8

Inspect the fitted object

print(fit)
## -------------------------------------------------------------
## RLT Classification Forest
## -------------------------------------------------------------
##               (N, P) = (80, 10) [5 continuous, 5 categorical]
##           # of trees = 200
##         (mtry, nmin) = (5, 5)
##       split generate = Best
##             sampling = 80% w/ replace
##           importance = permutation
##       OOB misclass = 27.5%
## -------------------------------------------------------------