Skip to main content

pydaltoolbox

PyPI version Python versions License: MIT

Python port of the R DAL Toolbox — leveraging Experiment Lines for Data Analytics.

pydaltoolbox organizes data analytics workflows (preprocessing, modeling, evaluation, tuning, extensibility) around a consistent Experiment Line style API. Its value is not reimplementing classic algorithms — it reuses scikit-learn, numpy and pandas — but a rigid, extensible interface that wraps them with a uniform fit / action / evaluate contract. Building a new method means subclassing a base and implementing two or three methods; sampling, evaluation and tuning come for free.

It is the foundation of the DAL Python family that also includes pytspredit (time series prediction) and pyharbinger (time series event detection).


Installation

Install the latest release from PyPI:

pip install pydaltoolbox

Beta: the v1 foundation is released and new features are landing incrementally (one release at a time) on the way to the 1.0.0 launch. The badge above always shows the latest version. Feedback and issues are very welcome.

The development version is available on GitHub:

pip install git+https://github.com/cefet-rj-dal/pydaltoolbox.git

For a local, editable clone (for contributing or running the examples):

git clone https://github.com/cefet-rj-dal/pydaltoolbox.git
cd pydaltoolbox
pip install -e ".[dev]"
pytest

First experiment

import pydaltoolbox as dal

iris = dal.datasets.iris()
slevels = list(iris["Species"].cat.categories)

split = dal.SampleRandom(seed=42).train_test(iris)
train, test = split["train"], split["test"]

model = dal.ClaDtree("Species", slevels, random_state=42).fit(train)

pred = model.predict(test.drop(columns="Species"))   # pass only the features
ev = model.evaluate(test["Species"], pred)
print(ev["metrics"])

Documentation

  • Processed examples — notebooks executed end to end, with real output embedded (the "processed documentation", analogous to the R package's examples/).
  • develop/planejamento.md — the migration roadmap (phases, scope, status).
  • develop/deploy.md — step-by-step guide to publish the package to PyPI.

The example sources live under notebooks/ (jupytext Markdown) and are rendered into examples/ by:

pip install -e ".[docs]"
python tools/build_examples.py

Status

Beta — the v1 foundation is complete and Cycle 1 is extending coverage toward 1.0.0 (see the roadmap for the live plan and version map). Implemented:

  • Core Experiment Line: DalBase, DalTransform, DalLearner, Predictor, set_params, and the adjust_* helpers.
  • Sampling: SampleRandom, SampleStratified (train_test, k_fold), train_test_from_folds.
  • Classification: Classification base + ClaDtree, ClaRpart, ClaKnn, ClaNb, ClaRf, ClaBagging, ClaBoosting, ClaSvm, ClaMlp, ClaGlm, ClaMultinom, ClaGlmnet, ClaMajority; ClaXgboost (gradient boosting) needs the optional [xgboost] extra (pip install pydaltoolbox[xgboost]) (0.13.0).
  • Regression: Regression base + RegLm, RegDtree, RegKnn, RegRf, RegSvm, RegMlp.
  • Clustering: Clusterer base + metrics + ClusterKmeans, ClusterDbscan, ClusterGmm, ClusterHclust, ClusterPam, ClusterCmeans (0.6.0).
  • Transformations:
    • normalization: MinMax, ZScore
    • cleaning: NaRemoval, ImputationSimple, OutliersBoxplot, OutliersGaussian
    • predictive imputation: ImputationPredictive, ImputationTree (0.3.0)
    • encoding / projection: CategMapping, DtPca
    • feature selection: FeatureSelectionCorr, Lasso, InfoGain, Fss, Stepwise, Relief
    • feature engineering: FeatureGeneration, Aggregation, HierarchyCut (0.4.0)
    • smoothing: SmoothingInter, SmoothingFreq, SmoothingQuantization, SmoothingCluster (0.2.0)
    • balancing: SampleBalance, BalSubsampling, BalOversampling
    • sampling: SampleSimple, SampleGroups (0.5.0)
    • curvature helpers: FitCurvatureMin, FitCurvatureMax
  • Pattern mining: PatternMiner base + PatUtils + PatApriori (frequent itemsets and association rules) (0.9.0); rule filters RuleFilter / RuleFilterNone / RuleFilterDara + the dara divergent-association ranking (0.10.0); RuleFilterInterest and support/confidence auto-estimation SupportThreshold / ConfidenceThreshold (0.11.0); PatPrefixSpan sequential pattern mining (native PrefixSpan) (0.12.0). PatApriori and the rule filters need the optional [patterns] extra (pip install pydaltoolbox[patterns], brings mlxtend); PatPrefixSpan is pure numpy/pandas.
  • Tuning: DalTune base + ClaTune, RegTune, CluTune.
  • Autoencoder bases: AutoencBaseE, AutoencBaseEd (identity, for extension).
  • Reference datasets: datasets.iris(), datasets.diabetes(), datasets.boston().

Wrappers are thin adapters over scikit-learn / numpy / pandas, each reproducing the R backend's default behavior (e.g. SVM standardizes inputs like e1071, trees pre-prune like tree/rpart). Pattern mining is complete (Cycle 1 done) and ClaXgboost adds the [xgboost] extra; see the roadmap for what is still coming — the remaining dependency-gated extra (graphics [viz]).


License

MIT

Metadata

Release files for pydaltoolbox 0.13.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for pydaltoolbox 0.13.0
File Size Uploaded Provenance Actions
pydaltoolbox-0.13.0.tar.gz 72.2 kB Without provenance Details for pydaltoolbox-0.13.0.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for pydaltoolbox 0.13.0
File Interpreter ABI Platform Actions
pydaltoolbox-0.13.0-py3-none-any.whl Python 3 none any Details for pydaltoolbox-0.13.0-py3-none-any.whl

Total release size: 173.8 kB

Release files / pydaltoolbox-0.13.0.tar.gz

Download URL pydaltoolbox-0.13.0.tar.gz
Size 72.2 kB
Tags Source
SHA-256 checksum
How to use checksums
5df20c2c3f77d8fbf97a2eb5f53af25ef1833848c64fbff49864f91d7ddc9757
BLAKE2b-256 checksum
How to use checksums
c463c6cd6d489a033b1ec7144713efee7cef876c0cd4c430d33f30d52d3b0d77
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.14.0

Release files / pydaltoolbox-0.13.0-py3-none-any.whl

Download URL pydaltoolbox-0.13.0-py3-none-any.whl
Size 101.6 kB
Tags Python 3
SHA-256 checksum
How to use checksums
aa6de027bf1d0f69bbc50205098c035c04baf4a1daa91e403de63af6c8bc97a7
BLAKE2b-256 checksum
How to use checksums
8df25d75e0fd5d90f08914a448c11b0b3be114a51ebf81343c66eff27d5e10fd
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.14.0

Security for pydaltoolbox 0.13.0

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page