文档下载 文档下载导航
设为首页 | 加入收藏
搜索 请输入内容:  
 导航当前位置: 文档下载 > 所有分类 > FACTORIAL DESIGNS
侵权投诉

FACTORIAL DESIGNS

KEYFINDER is a menu-driven interactive program that assists statisticians in the difficult and time consuming tasks of generating, randomizing and tabulating

KEYFINDER - A COMPLETE TOOLKIT FOR GENERATING

BLOCKED AND/OR FRACTIONAL-REPLICATE FACTORIAL DESIGNS

VERSION 3.3 - OVERVIEW

1. Introduction

KEYFINDER is a menu-driven interactive program that assists statisticians in the difficult and time consuming tasks of generating, randomizing and tabulating factorial designs in completely general situations. Its particular forte is the use of search procedures to generate blocked and/or fractional-replicate designs with user-specified confounding and aliasing properties. At present, there is no other system on the market with comparable capabilities.

To obtain a free copy of the KEYFINDER program and manual, simply send an email request to the author. Full details are given at the end of this overview. KEYFINDER is distributed as an executable program file and runs on Windows 95/98/NT/2000, SunOS, Solaris and Silicon Graphics machines. The source code was written in Prolog

Chapters 2 and 3 below give a general overview of the facilities available in KEYFINDER. For detailed instructions on how to use the program to generate experimental designs, you should consult the comprehensive KEYFINDER manual. The latter document also includes a theoretical treatment of the basic principles underlying KEYFINDER.

KEYFINDER is a system which is intended to help statisticians with those parts of experimental design construction that are difficult or tedious. It is not a system for those with little or no statistical knowledge. A basic understanding of the principles and underlying assumptions of experimental design and analysis of variance is assumed throughout this document and in the KEYFINDER manual.

A list of KEYFINDER publications follows Chapter 3.

2. Design Generation

KEYFINDER generates factorial designs involving a number of treatment factors (denoted A, B, ...), each of which is tested at a number of distinct "levels". KEYFINDER can generate

multiple, single or fractional replicates In multiple-replicate designs, each combination of treatment levels is tested several times, in single-replicate designs just once. In fractional-replicate designs, only a subset of the possible combinations of treatment levels is tested. Designs may be

symmetrical or asymmetrical

A symmetrical design is one in which all the block and treatment factors have the same number of levels; in an asymmetrical design, the numbers of levels may differ.

Designs may be generated by a number of methods, the most important being design keys

The design key was originally conceived as a device both for constructing fractional-replicate and/or blocked factorial designs and for identifying the aliasing and/or confounding patterns therein (Patterson, 1965, 1976; Patterson and Bailey, 1978). KEYFINDER reverses this process by searching for design keys yielding designs matching the user's a priori aliasing and confounding requirements (Zemroch, Lunn, Baines and Clithero, 1989). In particular, KEYFINDER can generate designs of fixed resolution

guaranteeing that only interactions above a given order are aliased with main effects (Box and Hunter, 1961) and also compromise designs

which additionally protect user-defined lists of higher-order interactions against aliasing. The experimental units in all the design types described above can be arranged into blocks. The block factors, denoted P, Q, ... may be

第1页

你可能喜欢

热门文档

相关文档

站点地图 | 文档上传 | 侵权投诉 | 手机版
本站所有资源均来自互联网,本站只负责收集和整理,均不承担任何法律责任,如有侵权等其它行为请联系我们.
文档下载 Copyright 2013 doc.guandang.net All Rights Reserved.  -  email