Modeling Manufacturing Data with a Doubly Truncated Logistic

by Mary L. Baggett .

Abstract: Statistical tolerance analysis typically assumes data from manufacturing processes have a normal distribution. Processes from a wide variety of industries have output with skewed distributions or finite ranges. One method of coping with this problem has been to use the beta distribution to model manufacturing data. The use of a doubly truncated logistic distribution for this situation is examined. This distribution is suitable for a wider class of processes than the beta distribution. Parameter estimation for this distribution is more straightforward than it would be for a truncated normal distribution. Comparisons are made between the use of the normal and truncated logistic distributions for predicting the proportion of product that fails to meet requirements using actual manufacturing data.

Key Words: Truncated Logistic Distribution, Quality Improvement, Manufacturing Data

Authors:
Mary L. Baggett, mbaggett@olemiss.edu

Editor: Ramon V. Leon , rleon@utk.edu

READING THE ARTICLE: You can read the article in portable document (.pdf) format (161843 bytes.)

NOTE: The content of this article is the intellectual property of the authors, who retains all rights to future publication.

This page has been accessed 2291 times since July 24, 2006.


Return to the InterStat Home Page.