I want to compare the different learning to rank techniques statistically. For that, I am using the P-Values and ANOVA. What are other methods? How to define this statistical confidence for better model
P-values won't give you a measure of effect size.
It is a little difficult to understand exactly what you want, but it may be that you need to use something like the AIC which is used to compare statistical models.
(ps- I recommend avoiding the term "statistical significance" as it can be misleading)
As Jhon said, you could use AIC or Bayesian Information Criterion (BIC) to compare statistical models,
Yes, AIC , BIC or p_value also MSE of the model are very statistical cretieria to compare between models or methods estimation.
with best regards
Dear Ashwini
The test of hypothesis is a branch of inference statistics can be used to compare many groups statistically, and there are many statistical tests and each one is suitable for a specific hypothesis which depends on the data type and its distribution.
Good Luck
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics