Rob Wiederstein
33d0fc31c7
End-to-end LightGBM fraud detection pipeline built as an R package, orchestrated by targets with data stored in MinIO via Apache Arrow. Includes 6-layer Lakehouse architecture, class imbalance tournament, formally tuned hyperparameters (PR-AUC 0.198), and Quarto RevealJS slides. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
35 lines
1.1 KiB
R
35 lines
1.1 KiB
R
% Generated by roxygen2: do not edit by hand
|
|
% Please edit documentation in R/functions.R
|
|
\name{run_imbalance_tournament}
|
|
\alias{run_imbalance_tournament}
|
|
\title{Run Class Imbalance Tournament}
|
|
\usage{
|
|
run_imbalance_tournament(
|
|
tasks,
|
|
windows,
|
|
feature_prefix,
|
|
bucket_name = "baf-fraud",
|
|
inputs_prefix = "05_model_input"
|
|
)
|
|
}
|
|
\arguments{
|
|
\item{tasks}{A tibble containing recipe_name, data_folder, and scale_pos_weight.}
|
|
|
|
\item{windows}{A tibble containing window_id, train_months, and test_month.}
|
|
|
|
\item{feature_prefix}{Character. The upstream dependency prefix (used to force DAG execution).}
|
|
|
|
\item{bucket_name}{Character. Bucket name. Default "baf-fraud".}
|
|
|
|
\item{inputs_prefix}{Character. The folder containing the sampled data. Default "05_model_input".}
|
|
}
|
|
\value{
|
|
A tibble with the summarized tournament results.
|
|
}
|
|
\description{
|
|
Trains LightGBM models across different class imbalance strategies
|
|
(Standard, SMOTE, Adasyn, etc.) using sliding time windows. Evaluates
|
|
performance using PR-AUC and calculates statistical significance.
|
|
Includes common-sense hyperparameter defaults to prevent overfitting.
|
|
}
|