Azure ML Lab Midterm 1 of CalStateLA- ABEL

March 8, 2019
Improved Performance of AUC
I improved this by adding a clip value and setting its “set of thresholds” to ClipPeaksAndSubpeaks and its “threshold” to percentile instead of leaving it at constant. Still in the clip value I set the “percentile number of upper thresholds to 90 and the “percentile number of lower thresholds” to -1. Then I selected the missing fro the “substitute value for peaks”. Then I added indicator columns within instead of selecting to overwrite the flag. The clip value was used to clip peaks and subpeaks to constrain the clomun values where the following was selected. Then in the split data, I changed the fraction of rows in the first output dataset from 0.5 to 0.99 and the random seed to 0 and kept everything else in the split data the way it was. I also used the edit metadata and had the data type to “unchanged” and the categories to “make categorical” then the fields to “unchanged”. These are all the changes I have made to improve the performance of this assignment.