User Guide

Operations Research Models & Methods (ORMM) is inspired by Paul A. Jensen’s Excel Add-ins. His Excel packages were last updated in 2011, and while I believe they do still work (for the most part), his work may become outdated in a couple of ways:

  • Excel is not as commonly used for OR, except in settings where security is of the utmost concern and/or modern languages like Python, R, Julia, C, C++, MATLAB, AMPL, or other modeling software are not available.

  • From what I understand, Microsoft has been trying to phase out VBA and move to Javascript. If this happens, this could significantly impact whether or not his packages will work.

  • His website and packages used to be available here, but currently I at least have not been able to load this webpage anymore - I’m not sure if UTexas took it down or not.

This python package aims to accomplish some of the same goals as Paul Jensen’s website and add-ins did, mainly to

  1. Be an educational tool that shows how abstract models (linear programs, integer programs, nonlinear programs, etc.) can be applied to real-life scenarios to solve complex problems.

  2. Help the practitioner by providing modeling frameworks, methods for solving these models, and problem classes so a user can more easily see how they may be able to frame their business problem/objective through the lens of Operations Research.

This repository contains subpackages for grouping the different types of OR Models & Methods. Currently this subpackage list includes

  1. mathprog: A subpackage for mathematical programs, including linear programs, mixed integer linear programs, nonlinear programs, and stochastic programs. Note for this subpackage that models and methods are not necessarily implemented in their abstract form, like Paul Jensen did - there are many python libraries that accomplish this task far better than I could (Pyomo, PuLP, GLPK to name a few). Thus, this subpackage here is dedicated to providing many problem classes, which show how these can be applied to real-life problems and provide an abstract/concrete model for that particular class of problems. Note that the abstract models can be built upon based on a unique business problem that may have more or fewer constraints, or a more complex objective to maximize/minimize.

  2. markov: A subpackage for discrete state markov analysis. Currently this only has implementations for discrete time markov processes, but continous time will be added in the near future. This includes the main function markov_analysis, which returns a dictionary of the results, as well as a print_markov function. The main method requires a transition matrix, but can then run simulations, analyze steady state and transient probabilities, and run cost analyses if additional arguments are passed.

Installation

$ pip install ormm

Examples

The mathprog subpackage has multiple problem classes, as well as functions for printing the solution of a solved concrete model and for returning a pandas dataframe containing information for sensitivity analysis. Following are some examples of a few of these problem classes.

  1. Resource Allocation: Optimize using scarce resources for valued activities.

from ormm.mathprog import resource_allocation
model = resource_allocation()
  1. Blending Problem: Optimize the mixing of ingredients to satisfy requirements while minimizing cost.

from ormm.mathprog import blending
model = blending()
  1. Employee Scheduling: Minimize the number of workers hired while meeting the minimum number of workers required for each period.

from ormm.mathprog import scheduling
model = scheduling(prob_class="employee")
  1. Rental Scheduling: Minimize the cost of the plans purchased (which rent units for different amounts of time) while satisfying the number of units needed for each period.

from ormm.mathprog import scheduling
model = scheduling(prob_class="rental")

For more details on optional parameters and usage, see the API Library Reference. For more details on the MathProg problem descriptions, see the Mathematical Programming.

Developer Environment

To use the same packages used in development (for creating additions / modifications), you may use the bash command below to install the dev requirements (recommended to do this in your virtualenv). This includes being able to run tests and add to the documentation.

$ pip install -e .[dev]