Markov Chain Monte Carlo Simulation Algorithms for quick convergence solution using solution engine of IBM SPSS Modeler - BC-438

Desired discipline(s): Mathematics, Mathematical Sciences, Statistics / Actuarial sciences
Company: Anonymous
Project Length: 6 months to 1 year
Preferred start date: As soon as possible.
Language requirement: English
Location(s): Vanderhoof, BC, Canada
No. of positions: 1
Preferred institutions: Thompson Rivers University, University of Northern British Columbia

About the company: 

Partner is IBM Software Business Partner in Canada

Please describe the project.: 

The project focuses on creating an API in Python that when compiled can use “hooks” within IBM SPSS Modeler to solve a modeling problem that uses Markov (or Semi-Markov) Chain Monte Carlo (MCMC) algorithm. The research objectives are to create sample problems and solutions that utilize MCMC, evaluate the two industry standard algorithms with existing limitations, write a Python language program for solution and compile an API that contains a data handler that interfaces with IBM SPSS Modeler for solution and visualization of obtained results.

Objectives:

  • Sample problems and solutions utilising MCMC
  • Evaluate existing industry standard algorithms
  • Compile and data handling API written in Python language
  • Interface with IBM SPSS Modeler engine for solution

Proposed methodology:

  • Categaorization and Evaluation of existing algorithms
  • Analysis of error bounds of current algorithms
  • Unknown methodology to be documented after research program

Required expertise/skills: 

• Minimum M.Sc. level in mathematical Statistics
• Fluency in Python in Eclipse Integrated Deevelopment Environment (IDE)