Rho andf LK Values

[Robbie]

I want to better understand the Rho and LK values - I have analysed data (2 subpopulation mixed weibull), which yield Rho of 0.9197 and LK value of -65.1679.

How do I properly interpret and use this infomation?

[David]

Hi Robbie,

Rho is the value of the correlation coefficient and really only applies when using rank regression. A discussion on the correlation coefficient can be found at http://www.weibull.com/LifeDataWeb/least_squares.htm under the Correlation Coefficient section.

The LK Value is the log-likelihood value and it applies when analyzing the data using maximum likelihood estimation (MLE). The LK Value by itself does not mean much. However, for a given data set, the LK Value can be compared across distributions and the distribution with a larger LK Value (less negative) indicates a better fit. Keep in mind that this only applies to MLE. Comparing the LK Value across different data sets is not valid. Additional information on MLE and the LK Value can be found at http://www.weibull.com/LifeDataWeb/appendix_a_parameter_estimation.htm#mle.

An overview of the parameter estimation methods are presented at http://www.weibull.com/LifeDataWeb/appendix_a_parameter_estimation.htm.

I hope this helps.

[Robbie]

Thanks David, thats clarified it for me.

Robbie