View Full Version : Value on the confidence bound?
vnigam
December 21st, 2007, 11:12 AM
Hi,
I have a distribution for a set of test data. I want to read the corresponding lower confidence value for 10% unreliability. How do I do that? Moreover, is there a way of knowing what were the estimated +/- on the estimated parameters for the selected Confidence?
Thanks
VN
Pantelis
December 22nd, 2007, 09:02 PM
Not sure of the question.... How to read? (read from where?)
or did you mean compute/estimate
vnigam
January 2nd, 2008, 10:53 AM
I have a plot of the distribution, linear fit and the lower and upper confidence bounds. This plot is Probability of failure Vs time. I want to read from this plot, the time at 10% probability of faliture on the fitted line and at the two bounds shown on the plot. I can appoximate it by eyeballing but, is there a way by which I could read the exact values? Hope I'm clear this time?
Thanks
David
January 2nd, 2008, 05:24 PM
How were the confidence bounds on the plot generated? There are multiple methods to calculate the confidence bounds. Two common methods are Fisher Matrix and Likelihood Ratio.
Fisher Matrix: http://www.weibull.com/LifeDataWeb/fisher_matrix_confidence_bounds.htm
Likelihood Ratio: http://www.weibull.com/LifeDataWeb/likelihood_ratio_confidence_bounds.htm
Additional information on confidence bounds can be found at http://www.weibull.com/LifeDataWeb/confidence_bounds.htm.
I hope this helps.
vnigam
January 3rd, 2008, 11:57 AM
David,
When I run the failure data using weibull++, we get a linear plot with the selected confidence bounds. Say the fit is a weibul fit with a beta and eta value is displayed. In order to generate a lower and upper CB it would internally estimate the bounds of the parameters for the selected method and CI right? What I see from the plot is the eta and beta values, and bounds about the linear fit showing the range in which the probability of failure may fall at a given time. I want to know if the software can give me the numeric estimates of the range or do I have to approximate by eyeballing it?
I hope I am clear this time?
Thanks
VN
David
January 3rd, 2008, 12:11 PM
OK. I think I got it this time. You can calculate exact values using the Quick Calculation Pad (QCP). Once the data has been calculated, you can access the QCP from the data sheet by selecting Quick Calculation Pad from the Data menu or by clicking on its icon on the Control Panel to the right of the data sheet. From a plot sheet, you can access the QCP by clicking on its icon on the Control Panel. From the Basic Calculations tab, you can generate results from the probability plot by selecting Std. Prob. Calculations under Options for Calculations. You can specify Results as Reliability or Results as Probability of Failure under Results Options. On the Confidence Bounds tab you can specify options to display the confidence bouds. Just make sure you have one-sided or two-sided selected as you have on the plot. Back on the Basic Calculations tab, enter the Mission End Time and click on Calculate to return the results. Additional information on the QCP can be found in the Weibull++ 7 User's Guide, pages 263-275.
I hope this helps and I apologize for the confusion.
vnigam
January 4th, 2008, 08:03 AM
Thanks a lot for the explaination.
regards VN
vBulletin® v3.8.4, Copyright ©2000-2012, Jelsoft Enterprises Ltd.