- Title
- Sampling behaviour of estimators for the generalized lambda distribution using distributed computing
- Creator
- King, Robert Arthur
- Relation
- 8th International Conference on Fixed Point Theory and its Applications (IASC). Proceedings: IASC 2008: Joint Meeting of 4th World Conference of the IASC and 6th Conference of the Asian Regional Section of the IASC on Computational Statistics & Data Analysis (Yokohama, Japan 5-8 December, 2008) p. 876-879
- Relation
- http://math.science.cmu.ac.th/ICFPTA2007/proceeding.php
- Publisher
- Japanese Society of Computational Statistics
- Resource Type
- conference paper
- Date
- 2008
- Description
- The generalized λ distribution (gλd) (Ramberg & Schmeiser 1974, Freimer et al. 1988) is a flexibly-shaped distribution that can take on a very wide variety of shapes in the one distributional form. A large number of different estimation methods have been proposed for the distribution (see (Su 2007, Fournier et al. 2007) for recent examples). Many of these estimators do not automatically estimate standard errors for the estimates produced, usually suggesting a parametric bootstrap to calculate them. I present a simple distributed computing framework for these parametric bootstrap calculations so that users of these estimation methods can benefit from the work of other users. The framework consists of a web-facing database, and R code to connect to the database. When users carry out a parametric bootstrap for the estimated parameters in their own problem, they are invited to share the results to the database, so that later users can make use of those results without the computational expense of doing the simulation. This differs from large scale distribution programs in that the server is not allocating the parameter space to nodes, but by making use of existing simulations, concentrates on those areas of the parameter space most used in practice. I illustrate the framework using the starship estimation method (King & MacGillivray 1999).
- Subject
- generalized lambda distribution; distributed computing; sampling distribution; simulation
- Identifier
- uon:6145
- Identifier
- http://hdl.handle.net/1959.13/802621
- Identifier
- ISBN:9784990444518
- Reviewed
- Hits: 1221
- Visitors: 1412
- Downloads: 0
| Thumbnail | File | Description | Size | Format |
|---|