When we analysis the sensitivity of various input parameters in final models evolved by for example ANN, genetic programming, Linear regression or other models, the most important parameter may be different in different models.
is this logical or not?
regards
Ali
Madhavi Latha Gali
Results of sensitivity analyses of two different ANN models can also be different. Hence it is quite logical to get different results for ANN, GP and regression models.
2 votes
1 thanks
Marek Wojciech Gutowski
Practically no computer generated result is perfect, at least in real numbers. Soft computing is no exception. Thus something must be wrong if results obtained by different methods differ too much. Roughly, when their uncertainty intervals do not intersect.
0 votes
0 thanks
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Graphs
More Ali Pourzangbar's questions See All
Similar questions and discussions