minnetonka 0.0.1

Motivation

Over the last 25 years, I built many business simulation models, initially in iThink, then in Powersim, and since 2006 in Forio SimLang. Forio SimLang is great for business modeling, powerful and expressive, with great support for arrays. But lately I have encountered several occasions in which SimLang is a good solution for modeling part of the problem, and an existing Python package is a good solution for modeling another part of the problem.

For example, in 2016 I built a model in SimLang that included some simple graph logic. There are no graph primitives in SimLang, so I created the graphs using the primitives available: arrays and enums and floats. There are no built-in analytics to find the shortest path in a weighted graph, so I wrote one, using the Floyd-Warshall algorithm.

That model would have been far simpler had I used the Python package NetworkX for the graph logic, and used SimLang for the other 80% of the model. What I needed was the expressivity of SimLang, within a Python package, so I could easily integrate with NetworkX.

Minnetonka is that combination: the power of Forio SimLang for business modeling, delivered as a Python package. Minnetonka is an appropriate solution when you want to model a business (or other social organization) in Python.

Features

A Minnetonka model is a collection of variables. Each variable takes a value, a value that can be of any Python data type: an integer, a float, a tuple, an array, a dict, etc.

A Minnetonka model is simulated over time. The variables in a model take a succession of values during the simulation.

Minnetonka variables are defined in terms of other Minnetonka variables via Python functions, allowing arbitary Python code to be executed at every simulation time step, for every variable.

Minnetonka supports stocks and flows. Stocks allow circular dependencies among variables, to model the circular causality underlying many business situations.

Minnetonka introduces treatments, a primitive for value modeling. A single variable can take one value in one treatment and take a different value in a different treatment. For example, business earnings might be $20 million per year in the as-is treatment, and $25 million per year in a to-be treatment, with a planned investment generating additional earnings.

Getting Started

• Building a model in Minnetonka

• API reference

Installation

Dependendencies

Minnetonka requires Python 3.6, and depends on NumPy and SciPy.

Contact

Send me an email.

License

Apache License, Version 2.

Naming

Minnetonka is named after Lake Minnetonka, the ninth largest lake in the US state of Minnesota.