R Observatory: Metrics for evoFE version 1.0.0

Balamuta, James Joseph

doi:10.5281/zenodo.21843040

evoFE

1.0.0Package record

Evolutionary Feature Engineering

0packages depend

1.2Kdownloads / year

83.7%test coverage

13/13flavors OK

Overview

About

Maintained by Gustavo PereiraCRAN page ↗GitHub ↗Issues ↗

Automates feature engineering using evolutionary algorithms inspired by genetic programming. Starting from raw input features, the package evolves candidate transformation recipes through selection, crossover, and mutation, evaluating fitness via cross-validation or train/validation splits with gradient-boosted tree models ('LightGBM' or 'XGBoost'). Built-in transformers include arithmetic, logarithmic, and power operations, interaction terms, target encoding, quantile and log-based binning, principal component analysis, truncated singular value decomposition, Uniform Manifold Approximation and Projection (UMAP) dimensionality reduction, and minimum spanning tree (MST) graph-based clustering. The evolutionary search yields an optimised feature recipe that can be applied to new data for prediction. Methods are described in McInnes et al. (2018) doi:10.21105/joss.00861, Ke et al. (2017) doi:10.48550/arXiv.1711.08789, Chen and Guestrin (2016) doi:10.1145/2939672.2939785, Gagolewski (2021) doi:10.1016/j.softx.2021.100722, Gagolewski (2026) doi:10.32614/CRAN.package.lumbermark, and Gagolewski (2026) doi:10.32614/CRAN.package.deadwood.

Changes in 1.0.0

New Features

**Untouched confirmation holdout (holdout_frac)**: With evaluation_strategy = "cv", a stratified fraction of rows is excluded from the entire search and scored once after evolution. The result is reported as holdout_fitness and a new search_gap diagnostic (holdout minus validation score) on the evo_recipe object and in summary(), making search-level overfitting visible.
**Leakage-safe CV fold strategies (cv_strategy)**: New options "time" (contiguous chronological blocks ordered by time_col, so validation always lies in the future of training) and "group" (all rows of a group_col entity stay in one fold, balanced greedy assignment) alongside the default "random".
**Multi-fidelity evaluation (multi_fidelity, mf_sample_frac, mf_warmup_frac)**: Optional cost reduction in which warm-up generations screen individuals on row-subsampled folds; the most promising half is re-evaluated at full fidelity before any selection decision, keeping fitness comparisons apples-to-apples.
**New transformer date_diff**: Signed difference in days between two datetime columns.

Breaking Changes

The mbo_infill_opt = "ea" option is deprecated and ignored with a warning. The emoa package is no longer needed.

Enhancements

Migrated Bayesian Optimisation backend from mlrMBO / ParamHelpers / smoof / lhs to mlr3mbo / paradox / bbotk. The public API (make_tunable(), lightgbm_mbo, xgboost_mbo) remains fully unchanged.
Removed legacy packages mlr, randomForest, DiceKriging, and emoa from Suggested dependencies. The manual Kriging-to-RandomForest surrogate fallback has been replaced by mlr3mbo's transparent internal surrogate management.
Reduced package dependencies in Imports from 12 to 11.
Added support for a **Hybrid Island Population Model** in evolve_features() via parameters islands, migration_interval, migration_rate, and gene_migration_prob.

Show all changes in 1.0.0Hide the rest of the changes in 1.0.0

**Recipe-Level Migration**: Exchanges complete successful feature recipes between islands in a Ring topology to preserve co-adapted interactions. Guarded against edge cases where dynamic population sizes shrink below migration_rate.
**Gene-Level Migration (Injection)**: Periodically injects highly successful individual features (genes) from the ring neighbor's best individual into the local mutation pool. Genes diffuse naturally through the ring over successive migrations, creating a gradient of innovation where each island builds hierarchically on its neighbor's features.
**Independent Local Stagnation**: Dynamic population sizing and adaptive exploration rates are computed independently per island, allowing stagnant islands to expand and explore while active ones stay compact. Population growth is capped at 5x pop_size to prevent runaway expansion.
**Verbose Reporting**: Clear logging has been integrated to distinguish exactly which island and individual is being evaluated, including gene pool injection events.
Added support for a **Hybrid Feature Mask Approach** for raw feature selection alongside evolved feature engineering:
**Sigmoid Active Mask Initialization**: Guides active feature masks of new individuals using baseline model importances centered at average feature importance.
**Smart Mask Mutation Operators**: Supports multi-column toggles scaled via a dynamic geometric distribution, single raw feature toggles with temperature-scaled importance probability, and complete mask recalculation.
**With-Replacement Multivariate Sampling**: Employs with-replacement sampling and deduplication for multivariate transformers to concentrate inputs around active, important features.
**Pruning Safety Floor Restoration**: Safeguards evaluations by automatically restoring raw features to meet the min_active floor if all genes are pruned during evaluation.
**Cache Hit Logging Reduction**: Silenced redundant (Cache Hit) prints for centroid distance components to reduce console verbosity.

From NEWS.md in the release

Install

R (CRAN)pak (dev)Ubuntu (r2u)Fedora (COPR)openSUSE (OBS)

install.packages("evoFE", repos = "https://cloud.r-project.org/")

Installs the current CRAN release. Prebuilt binaries on Windows and macOS.

if (!requireNamespace("pak", quietly = TRUE))
  install.packages("pak", repos = "https://cloud.r-project.org/")
pak::pak("tanopereira/evoFE")

Development version from tanopereira/evoFE. Installs pak first if it is not already available.

Release

# enable cranapt (r2u) if it is not already set up, then install
grep -rqs r2u.stat.illinois.edu /etc/apt/sources.list /etc/apt/sources.list.d/ \
  || curl -fsSL https://raw.githubusercontent.com/eddelbuettel/r2u/master/inst/scripts/add_cranapt_noble.sh | sudo bash
sudo apt install -y r-cran-evofe

Fast Ubuntu binaries via r2u (no compilation). The cranapt setup runs once and is skipped when it is already enabled. We list this channel from recorded r2u download activity rather than from a live index of the repository, so treat it as last known rather than guaranteed.

bspm (used by r2u) also needs make: sudo apt install make

# enable the iucar/cran COPR if it is not already present, then install
ls /etc/yum.repos.d/*iucar*cran*.repo >/dev/null 2>&1 \
  || sudo dnf copr enable -y iucar/cran
sudo dnf install -y R-CRAN-evoFE

RPMs from the iucar/cran COPR, rebuilt from CRAN.

For install.packages() integration in R, also: sudo dnf install R-CoprManager

# add the autoCRAN OBS repo (Tumbleweed) if it is missing, then install
zypper -q lr | grep -qi autoCRAN \
  || sudo zypper addrepo https://download.opensuse.org/repositories/devel:/languages:/R:/autoCRAN/openSUSE_Tumbleweed/devel:languages:R:autoCRAN.repo
sudo zypper install -y R-evoFE

RPMs auto-built from CRAN on openSUSE OBS (Tumbleweed). We list this channel from recorded autoCRAN download activity rather than from a live index of the repository, so treat it as last known rather than guaranteed.

Health

CRAN checks

13OK

CRAN check page ↗

Slowest check: 11.4 min · r-release-linux-x86_64

Status changes

OK13 OK2026-08-05
NOTE12 OK, 1 NOTE2026-08-01
OK8 OK2026-06-10

Advisories

none known

How this is measured·How advisories are found

Documentation

In release 1.0.0

READMEYes · 552 wordsVignettesYes · 1 · dynamicNEWSYes · 100% structured · mentions 1.0.0Citation fileNo

Description cites a JOSS paper

Help pages: 48 (9 internal)includes a package overview page
Examples that run: 21%3 run in a plain check · 10 only when donttest examples are enabled (R CMD check --as-cran) · 1 never run
Exported functions with examples: 37%
Documented parameters: 100%
Return-value docs: 100%
Help pages citing references: 0%

Also includes 3 demo scripts (run with demo())

Vignettes (1)

Getting Started with evoFE HTML

How these are measured

Downloads

1.2K

CRAN downloads in the past year

Rank #14,253 · ~3/day · ~97/mo

Releases R versions

Daily download trend is not available in this view yet.

30930 days

82590 days

1.2K1 year

Compare downloads with other packages →

Also on32 r2u2 autocran

How downloads are counted

Repository

tanopereira/evofe ↗Account: tanopereira

6Stars

0Forks

0Open issues

0Open PRs

0Releases

290Commits

1Commit identities

290 commits · Last activity 2026-09-15 · +20% stars, 30d · Default branch at 1.1.0, ahead of CRAN 1.0.0 (checked 2026-10-18)

Responsiveness

not measured

Stars over time

2026-08-26 · 52026-10-05 · 6

How these are measured

PracticesBeta

4 development-tooling and community-health practices detected across 3 families in the upstream repository

Checks run against github.com/tanopereira/evofe on 2026-10-18.

Continuous integration (1)

GitHub Actions

Docs source (1)

GitHub Pages site last deployed 2026-09-15

CRAN release process (2)

cran-comments.mdCRAN-SUBMISSION

How this is detected·Detection ruleset v3 (2026-09-26)

Dependencies

Declared dependencies

26 external dependencies (excludes base and recommended)

Installing evoFE adds 50 packages (33 with compiled code, 47 MB of source tarballs). Heaviest: BH 14 MB, data.table 5.7 MB, spacefillr 3.1 MB.

Depends (0)

none

Imports (12)

data.table lightgbm xgbooststatsdigest uwot quitefastmst genieclust paradox bbotk mlr3mbo lhs

LinkingTo (0)

none

Suggests (15)

glmnet RhpcBLASctl testthat knitr rmarkdown lumbermark deadwood keras3 httpuv jsonlite farff ranger DiceKriging rgenoud mlr3learners

Enhances (0)

none

Dependency license breakdown →

Reverse dependencies

0direct

0indirect

Nothing depends on this yet.

How these are counted

Code & Tests

People & History

People (1)

Gustavo Pereira

Author, Maintainer since 0.1.0 (2026)

Releases (2)

Pick two releases to compare their code metrics.

1.0.0Latest
record
2026-08-25 · current release · diff ↗ ·
added 26 exports
compute_calibrated_mae, compute_calibrated_rmse, compute_complexity_penalty, compute_ts_refinement, ensemble_islands, evo_evaluators, get_in_neighbors, get_neighbors, make_tunable, migration_config, policy_gibbs_pull, policy_gibbs_push, policy_push_uniform, policy_tiered_admission, register_evaluator, register_transformer, resolve_migration_transactions, topology_complete, topology_custom, topology_grid, topology_hypercube, topology_ring, topology_tiered, topology_torus, tournament_select, view
0.1.0
record
2026-06-09
R
R 4.6.0 released · 2026-04-24

Package metadata

First published: 2026-06-09
Total releases: 2 / 1 yrs
License: MIT + file LICENSE OSI
Download size: 244 KB

Cite

Cite this package

Run in R for the authors' preferred citation:

citation("evoFE")

Format

Pereira, G. (2026). evoFE: Evolutionary Feature Engineering (Version 1.0.0) [Computer software]. https://doi.org/10.32614/CRAN.package.evoFE

This package has no citation file, so citation() returns this entry built from DESCRIPTION.

Cite the R Observatory

For a number measured here: a download total, a coverage figure, an archival date.

Format

APA

Balamuta, J. J. (2026). R Observatory: Metrics for evoFE version 1.0.0 [Data set]. HJJB, LLC. Data release v2026-10-21. https://doi.org/10.5281/zenodo.21843040

From data release v2026-10-21, which the citation names so these numbers can be found later. More on citing and the projects behind them.

Report a problem with this page →