Learn R Programming

mistat (version 0.999)

FLEXPROD: The Quinlan Experiment at Flex Products

Description

Flex Products is a subcontractor of General Motors, manufacturing mechanical speedometer cables. The basic cable design has not changed for fifteen years and General Motors had experienced many disappointing attempts at reducing the speedometer noise level.

Usage

data(FLEXPROD)

Arguments

source

Kenett, R. and Zacks, S. (1998) Modern Industrial Statistics: The Design and Control of Quality and Reliability. Duxbury Press.

Details

Problem Definition: the product under investigation is an extruded thermoplastic speedometer casing used to cover the mechanical speedometer cable on automobiles. Excessive shrinkage of the casing is causing noise in the mechanical speedometer cable assembly. Response variable: the performance characteristic in this problem is the post extrusion shrinkage of the casing. The percent shrinkage is obtained by measuring approximately 600mm of casing that has been properly conditioned $(A)$, placing that casing in a two hour heat soak in an air circulating oven, reconditioning the sample and measuring the length $(B)$. Shrinkage is computed as: $Shrinkage = 100 * (A-B)/A$. Factor Levels: Existing (1) - Changed (2) Number of Replications: four random samples of 600mm from the 3000 feet manufactured at each experimental run. Data Analysis: signal to noise ratios $(SN)$ are computed for each experimental run and analyzed using main effect plots and an ANOVA. Savings are derived from Loss function computations. The signal to noise formula used by Quinlan is: $$\eta = -10 log_{10} (\frac{1}{n} \sum y^2)$$

Examples

Run this code
data(FLEXPROD)

aov(SN ~ . , data=FLEXPROD)

Run the code above in your browser using DataLab