1 / 14

SHOJIMA Kojiro The National Center for University Entrance Examinations shojima@rd.dnc.ac.jp

Asymmetric von Mises Scaling. SHOJIMA Kojiro The National Center for University Entrance Examinations shojima@rd.dnc.ac.jp. Purpose of Research. Development of an asymmetric multidimensional scaling (MDS) method using a technique from directional statistics

lave
Télécharger la présentation

SHOJIMA Kojiro The National Center for University Entrance Examinations shojima@rd.dnc.ac.jp

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Asymmetric von Mises Scaling SHOJIMA Kojiro The National Center for University Entrance Examinations shojima@rd.dnc.ac.jp

  2. Purpose of Research • Development of an asymmetric multidimensional scaling (MDS) method using a technique from directional statistics • Asymmetric von Misesscaling (AMISESCAL)

  3. Directional Statistics (c.f., Mardia & Jupp, 2000) • A branch of statistics dealing with angles, courses, and directions as data • Magnetic field analysis, animalmigration, disease transmission route, etc.

  4. Slider von Mises distribution • Normal distribution in directional statistics μ: mean direction κ: concentration

  5. Model ||xi-xj|| xj δij θji Person j μj κj κi μi xi Person i πji=f(θji|μj, κj) θij πij=f(θij|μi,κi) 1 5

  6. Stress Function • Optimization • 1st Stage: Genetic Algorithm (GA) • 2nd Stage: Steepest Descent Method (SDM)

  7. 7 7 7 7 7 1 1 1 7 7 7 1 1 1 1 1 7 7 7 7 1 1 1 1 7 7 1 1 1 1 B C C A D A D B A C B

  8. Omnidirectionality and the Amount ofOne-sided Love Omnidirectional and more love Omnidirectional and less love Reduces to the conventional von Mises distribution when ω=1/(2π)

  9. Problem xj δij Person j xi πij=f(θij|μi,κi) Person i 1 5

  10. Stress Function (2) • Adding Penalty Function U • Reward when there are one-sided love targets in the direction of heavy density • Penalty when there is no target in the direction of heavy density • Optimization • GA+SDM

  11. Result 7 7 7 7 C C 1 1 7 1 1 7 1 1 D A A D B B

  12. Sociometric Data(Chino, 1997, p.13, Revised)

  13. Result

  14. Future Tasks • Dealing with diagonal elements • Expansion to 3D model space • Expansion to 2 mode (multi-group or longitudinal) data Thank you for your attention. Kojiro Shojima (shojima@rd.dnc.ac.jp)

More Related