In forecasting applications, there are different strategies for sampling based on the method used. In computer intelligence, scholars need to define a longer out-of-sample period and have both testing and validation periods. In the conventional forecasting (e.g. econometrics, time series analysis), we just define an out-of-sample period for both testing and validation.
- How do you think about different uses of out-of-sample period?
- How do you distinguish between testing and validation periods?
- How do you define testing and validation periods (e.g. which one should be longer and why)?
Okan Duru
However, it is almost essential in e.g. ANN to define a testing period as well as a validation period. Do you mean it is needless?
Let me give some details from web search that may help us.
- In the first part you just look at your models and select the best performing approach using the validation data (=validation)
- Then you estimate the accuracy of the selected approach (=test)
0 votes
0 thanks
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
More Okan Duru's questions See All
Similar questions and discussions