I would like to do polynominal regression on a set of data. Theoritically, it should be that the deeper the polynominal, the higher (R^2) I can get, but it is also known that overfitting could be a bigger problem if I go too deep. Then how should I do this polynominal regression? How can I determine how many items are there in such a task?
You can use the Akaike information citerion (AIC) or its small-sample version (AICc) to choose the degree of your polynomial term. With this method, the log of maximized likelihood (L) is penalized by the number of coefficients in the model.(k): AIC = -2 L + 2 k. The smaller the AIC, the better the model.
You can use the Akaike information citerion (AIC) or its small-sample version (AICc) to choose the degree of your polynomial term. With this method, the log of maximized likelihood (L) is penalized by the number of coefficients in the model.(k): AIC = -2 L + 2 k. The smaller the AIC, the better the model.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression