Replicate R and matlab results in finding the optimal threshold from ROC curve -


I am using optimum code in package R To find the optimum range of cutoff point from optimum ROC curve, maximize the index of criteria audience:

  J = sensitivity + specificity - 1  

I I am trying to do the matlab in the same function perfcurve I run perfcurve with the default criteria for the two axes, in the x coordinate Fpiar and Teepiar in y- coordinates. Returns a matrix with perfcurve threshold and chooses one of them according to the criteria.
The problem is that the optimal range providing matlab is not equal to R, however, according to optimal R, the threshold is included in the matrix which returns matlab.

How can I repeat the results, which is in return matlab with R? I suspect that the norms have not been properly set up in the matlab for the index of Euden.

If you see (especially the OPTROCPT line) You will see that the thread which Matlab uses to find the best threshold is quite different, and it also includes cost metrics in the optimization criteria.

If you have been done right in R, then use the X and y return values ​​to calculate the Uses index for each range , And then choose the best (see how to do this how to do it).


Comments

Popular posts from this blog

ios - How do I use CFArrayRef in Swift? -

eclipse plugin - Run java code error: Workspace is closed -

c - Error on building source code in VC 6 -