ProdPack

Efficiency measurement and productivity analysis are closely connected: a productivity study repeatedly estimates production frontiers, distances, and changes in operating performance. Maintaining them in separate packages made data preparation, model assumptions, result formats, and documentation harder to keep consistent.

DEAPack 2 provides one framework for:

• classical DEA efficiency models, including CCR, BCC, FDH, additive, SBM, EBM, and directional approaches;
• Malmquist-family, Luenberger, Hicks--Moorsteen, global, biennial, and environmental productivity analysis;
• economic, environmental, network, dynamic, panel, and metafrontier models;
• consistent data validation and structured score, target, slack, peer, component, and diagnostic results; and
• visualization, reporting, and reproducible research outputs.

The maintained project is available at:

• Package: DEAPack on PyPI
• Documentation: deapack.readthedocs.io
• Source and issues: github.com/daopingw/DEAPack
• Productivity methods: Productivity analysis guide
• Moving old code: DEAPack and ProdPack migration guide

Install the maintained package from PyPI:

python -m pip install --upgrade DEAPack

A current productivity workflow uses the same explicit data and result contracts as the rest of DEAPack:

from deapack import DEAData, FGNZMalmquist, dataset_info, load_dataset

dataset = "productivity_panel"
frame = load_dataset(dataset)
roles = dataset_info(dataset).roles

data = DEAData.from_frame(
    frame,
    dmu=roles["dmu"],
    period=roles["period"],
    inputs=roles["inputs"],
    outputs=roles["outputs"],
)

result = FGNZMalmquist().fit(data)
print(
    result.summary()[
        [
            "dmu_id",
            "period",
            "productivity_change",
            "efficiency_change",
            "technical_change",
            "score_valid",
        ]
    ]
)

See the DEAPack quickstart for the common data -> model.fit(data) -> result workflow and the method catalog for the maintained method inventory.

ProdPack remains here as a historical research-software record. Existing releases may still be used when reproducing an earlier analysis, but this repository is not the development home for new functionality or fixes.

DEAPack 2 is a redesign rather than a drop-in rename. Old scripts using mutable ProdNP objects, assigned x_vars/y_vars/b_vars, ref_type, and solve() should be migrated by first recording their empirical assumptions and then selecting the corresponding maintained DEAPack model. Do not assume that changing imports alone preserves the estimated quantity.

For migration questions, bug reports, or method proposals, please use the DEAPack issue tracker.