The "no multicollinearity" assumption is not needed. With perferctly correlated regressors (eg, when you leave all dummy variables in), the (unique) OLS estimate of any estimable function of the parameters is BLUE under the G-M conditions.
At 14:24, why DXB=0 can deduce DX=0? As B should be a column matrix, we cannot simply to make this deduction. Anyone can make some further explanation about this? Thanks.
while substituting the expression for beta hat at 10:47, why is there no x transpose? you initially wrote it correct and then ignored the transpose. Am I missing something here?
The "no multicollinearity" assumption is not needed. With perferctly correlated regressors (eg, when you leave all dummy variables in), the (unique) OLS estimate of any estimable function of the parameters is BLUE under the G-M conditions.
Well understood ,thank you so much
At 14:24, why DXB=0 can deduce DX=0? As B should be a column matrix, we cannot simply to make this deduction. Anyone can make some further explanation about this? Thanks.
while substituting the expression for beta hat at 10:47, why is there no x transpose? you initially wrote it correct and then ignored the transpose. Am I missing something here?
It's trivial if you make it transpose or not because eventually it gets cancelled out.
same question.
Thanks!
Nice explanation . thankyou.
Where the random sampling assumption at? :)
Thanks
Perfect explanation but it’s vry confusing n consideration part
senseless approach