It will help you to work with HDFS faster. ]]>

The line is:

upper.step.n = min(which(adjusted.knots > j))

Thanks for the post! I can't find references to isotonic regression for calibration using R anywhere...

I used your code as a base for mine, but found a bug in the fit.isoreg function and thought you should know about it:

In line 22 you wrote:

upper.step.n <- max(which(abs(adjusted.knots-j)==min(abs(adjusted.knots-j))))+1

This does not work, sometimes the lower step will be closer than the upper step and the value will be assigned to a step lower.

I it changed to:

upper.step.n j))

which always get the smallest adjusted.knot that is bigger than j.

Hope it helps!

Thanks again,

Diego

P.S. -- for some reason my first comment did not have the right correction. Sorry for the recomment.

]]>Thanks for the post! I can't find references to isotonic regression for calibration using R anywhere...

I used your code as a base for mine, but found a bug in the fit.isoreg function and thought you should know about it:

In line 22 you wrote:

upper.step.n <- max(which(abs(adjusted.knots-j)==min(abs(adjusted.knots-j))))+1

This does not work, sometimes the lower step will be closer than the upper step and the value will be assigned to a step lower.

I it changed to:

upper.step.n j))

which always get the smallest adjusted.knot that is bigger than j.

Hope it helps!

Thanks again,

Diego