Colin Smith

This library contains sipmath, a Python implementation of the SIPmath Modeler Tools for probabilistic modeling and forecasting by Probability Management Inc and pymetalog, a Python translation of RMetalog.

A SIP, or Stochastic Information Packet, is a method of representing a probability or frequency distribution as an array of samples and relevant metadata.

SIPmath is the process of performing aritmetic with SIPs directly. The SIPmath modeler tools allow for intuitive, sophistocated and powerful monte-carlo simulations on an enterprise scale.

SIPs and SIPmath can be used to create risk models that are

• Actionable: distributions can be easily manipulated arithmetically
• Additive: individual risk models can be easily aggregated into larger, comprehensive models.
• Auditable: the Open SIPmath Standard allows for the storage of unambiguous data with provenance intact.

The purpose of this repository is to bring the capabilities of the SIPmath Modeler Tools for Microsoft Excel to Python. Models created with this library are backwards compatible with the Excel tools, they can be exported to Excel and vice versa.

More information about SIPs and SIPMath are available WHERE

The metalog distribution is a family of continuous, univariate, and highly flexible probability distributions, parametrized by CDF data. Metalog distributions elegantly address the common need of finding a distributional representation of known data, especially when the data is not well represented by commonly used distributions.

Metalog Distributions can be used with SIPmath to create powerful risk models directly from data without known or pre-fit underlying distributions.

The native Python implementation of the Metalog Distributions in this Library, pymetalog, is a translation of RMetalog by Isaac Faber. More information is available in the paper published in Decision Analysis and the website.