Ch 13 page 21 ..... Full Normal Model and inference.... We assume ei follows normal distribution with mean 0 and Variance sigma square. This will help us to know the distribution of Yi and Bi. There is a note given in bottom of this page about to derive MLE for parameter "a and b". It is also mentioned that least square doesn't provide us the distribution of Yi. What does this mean....as it's already mentioned above that Yi is normally distributed with mean E(Yi)=a+bxi and Variance=sigma square Can anyone relate this ??
i think the note given in bottom tells that we can use MLE when the distribution is known but Least square estimator doesn't require distribution. As the dist. of Y is known so we can use MLE to estimate parameters. MLE require us to know the dist. rather than provide the dist. itself.