Member Retention Data Report to INCOSE

Abstract

Causal analysis is performed on membership data using characteristics of the data that is fed into three different algorithms, which use different methods to identify causality and are based on different assumptions. Depending on the amount of data and how strong is the causality, the methods can produce different results. Hence if different algorithms all show causality, it is highly likely that there actually is causality. For this reason four algorithms were run. They are called PC (for the inventors names), FGES (for fast greedy equivalent search), and FCI (for fast causal inference). FGES has an option to run with different criteria for scoring graphs, based on assumptions. We used two, one called DBIC (digital binary information criterion) and one called BDEU (Bayesian Dirichlet-likelihood equivalence and uniform).

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Dec 01, 2018
Accession Number
AD1084123

Entities

People

  • Michael D. Konrad
  • Sarah Sheard

Organizations

  • Carnegie Mellon University

Tags

Communities of Interest

  • Human Systems

DTIC Thesaurus Topics

  • Algorithms
  • Copyrights
  • Costa Rica
  • Data Sets
  • Department Of Defense
  • Engineering
  • Governments
  • Guarantees
  • Materials
  • Numerical Analysis
  • Performance Tests
  • Probability
  • Probability Distributions
  • Software Development
  • Students
  • Systems Engineering
  • Universities

Fields of Study

  • Computer science

Readers

  • Computer Science.
  • Neural Network Machine Learning.
  • Operations Research

Technology Areas

  • AI & ML
  • AI & ML - Bayesian Inference
  • AI & ML - Machine Learning Algorithms