Hello everyone,
I’m currently working on a Data Envelopment Analysis (DEA) project and I’m unsure whether to use original values or natural logarithms value for my input and output variables. I would greatly appreciate any insights on this topic.
Specifically, what are the advantages and disadvantages of each approach? Additionally, if anyone could recommend relevant studies or literature that discuss this issue, I would be very grateful.
Thank you in advance for your help!
Since DEA and FDH is nonparametric and non-linear, monotonic transformations like log etc. are usually not really meaningful when applied to all variables. However, rescaling individual variables could be justified when you e.g. reduce the effect of outliers.
Thank you for your insight! I would appreciate it if you could provide some references or research studies that support this viewpoint. It would be helpful to see how others have addressed the limitations of applying monotonic transformations, like logarithms, in nonparametric and non-linear methods (DEA). Looking forward to learning more about this.
- Badges
- Science topic