Information criteria | AIC | BIC | Uses and Differences.

Подписаться 106

Просмотров 1,4 тыс.

50% 1

Each model has:
a log likelihood (l)
a number of parameters (k)
a number of samples used for fitting (n)
AIC = 2k - 2l
Lower AIC via higher log likelihood or less parameters
BIC = ln(n)k - 2l
Lower BIC via higher log likelihood or less parameters or less samples used in fitting

Опубликовано:

20 сен 2024

Ссылка:

Скачать:

Готовим ссылку...

Добавить в:

Мой плейлист

Посмотреть позже

Комментарии : 11

@pradiptapattanayak8085 Год назад

good explanation

@machinelearningzone.6230 Год назад

Great that you liked it. There are other videos in the channel as well, you might like them. Consider liking, shring and subscribing if you find the content useful.

@sutannibhowmick7729 2 года назад

Liked the explanation!! Keep going!!

@machinelearningzone.6230 2 года назад

Thank you!!

@AyanGaanwala 2 года назад

Very informative..👏❤️👏❤️..

@machinelearningzone.6230 2 года назад

Thank you!!

@Meme_verse1 2 года назад

Very good and detailed

@machinelearningzone.6230 2 года назад

Thanks a lot! More such videos are upcoming.

@sniper4627 2 года назад

Hi there, you said higher value of (l) log likelihood means the data fits well But this video ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-4al2LfJz6Q8.html said the opposite The person in the video said the L hat or log-likelihood DECREASES as our model gets better. But you said the HIGHER the log-likelihood, the better Which one should I believe?

@machinelearningzone.6230 2 года назад

Our aim is to have a lower value of AIC/BIC .2k-2L being the formula,if L decreases with the model fitting better it would make the value af AIC /BIC greater ,thus would penalize a model which fits well which in turn would led us to choose an under fitting/model with high Bias. AIC/BIC helps us in selecting the simpler model which ha a low error. Hence L (log likelihood) decreases with a model fitting better.

@machinelearningzone.6230 2 года назад

Donot forget to like,share and subscribe if you liked the content..