DEAPack Documentation#

DEAPack is a Python framework for data envelopment analysis (DEA), efficiency measurement, productivity analysis, and environmental performance.

It gives researchers, students, and analysts a consistent way to prepare data, choose an appropriate model, fit it, and interpret the resulting scores, targets, slacks, and peers. It is especially useful when a study extends beyond classical DEA into economic, environmental, network, dynamic, or productivity analysis.

Note

DEAPack 2.0.1 is the current stable release of the redesigned 2.x API. If you are upgrading from DEAPack 0.1.x or ProdPack, begin with the Migrating from DEAPack 0.1.x and ProdPack guide.

Start here#

  1. Installation — install DEAPack and optional extras.

  2. Quickstart — fit a first model and inspect its result.

  3. Result contract — understand scores, targets, peers, validity, and diagnostics before reporting them.

Why DEAPack?#

A DEA study involves more than selecting a model acronym. The production technology, orientation, returns to scale, reference population, data roles, and treatment of undesirable outputs all affect the question being answered. DEAPack keeps those choices visible while providing a common workflow:

  • prepare a validated DEAData object;

  • fit a model through a consistent estimator interface;

  • inspect named result tables instead of unpacking solver arrays; and

  • create plots, reports, and reproducibility bundles from the fitted result.

You can therefore begin with a documented preset such as BCCInput and move to more specialized analyses without abandoning the same data and result conventions.

A first model#

from deapack import BCCInput, DEAData, load_dataset

frame = load_dataset("frontier_1x1")
data = DEAData.from_frame(
    frame,
    dmu="dmu",
    inputs="input",
    outputs="output",
)

result = BCCInput().fit(data)
result.summary()

This fits an input-oriented variable-returns-to-scale frontier. Continue with the Quickstart to inspect peers and targets, use your own data, and visualize the result.

What can I analyze?#

Area

Start with

Classical radial and non-radial DEA

Core efficiency models

Productivity change and scale

Productivity analysis and Scale, capacity, and decomposition

Cost, revenue, profit, and allocative efficiency

Prices and economic efficiency

Undesirable outputs and environmental performance

Environmental models

Network, dynamic, panel, and metafrontier structures

Network, panel, and dynamic models and Radial group and metafrontier analysis

Super-efficiency, cross-efficiency, diagnostics, and communication

Cross and super-efficiency evaluation, Share a result without losing its meaning, and Communicating performance results

The Method catalog is the authoritative inventory of methods available in the installed release. It distinguishes principal families, presets, and specialized methods without implying that planned work is already executable.

Documentation map#

Section

Use it for

Getting started

Installation, the first analysis, and migration from older releases

User guide

Data preparation, model discovery, result interpretation, reporting, and datasets

Models and analysis

Concepts, assumptions, equations, and method-specific behavior

API reference

Public classes, functions, parameters, and return contracts

Development

Architecture, extension points, contribution, and versioning policies

Citation and contributions#

Use Citing DEAPack when reporting a DEAPack computation, and cite the defining literature for the particular method used. Questions, reproducible bug reports, model proposals, and Documentation improvements are welcome through the project contribution guide.