Explain Akaike Information criteria.

922    Asked by IraJoshi in Data Science , Asked on Nov 30, 2019
Answered by Ira Joshi

Akaike Information Criteria or AIC, in simple words, is a trade-off between bias versus variance. It helps to calculate the relative quality of a statistical model for a given data. Mostly when AIC is compared between two models, the model with the least AIC is preferred.

AIC = -2*ln(L) + 2*k

L = Maximum value of Likelihood (log transformation applied for mathematical convenience)

k = Number of variables in the model

In the above equation, k is penalizing the over fitting phenomena of the model which means we can improve the training accuracy by incorporating insignificant variables in the model but in such case testing accuracy will decrease.

This technique can help as a regularization technique in logistic regression.



Your Answer

Interviews

Parent Categories