YOu can do a 2D interpolation as was done for the augmentation data. But I would suggest extremem circumspection in using this. You just don't have a lot of data points. It's hard to know what is a reasonable interpolation when you have values at just 0, 2 and 100. That's a big between.
Also, the range and spread of the C values is such that I would usually try to do a log transformation. But that is not possible here, as you start with a zero value. There may be other transformations that make for a more reasonable function. But even with such, your data is very sparse.
__________________
� � � � Tom Gutman
