14. Regression (cont.)

[@freeeagle6074]

Philippe has cleared many confusions that easily show up in other similar courses. For example, he's gone to details of whether we should keep t or t transpose and how that is related to whether we should keep xTy or yTx (10:05). He has dived into many similar details in each of other lectures. He has a strong empathy upon students' learning process. Philippe is a genius professor and a nice person.

[@aungkyaw9353]

Heaps of dot-connecting mental exercises for me. Thanks professor. Hopefully, your High-Dimensional Statistics course will be available on youtube as well.

[@吴舒晨]

It really feels like if you know more, you can get more from this class. Many thanks~

[@chuzai2156]

yeah, Brownian bridge, Gradient, it's not a beginner level class. But it's a good way help me to connect different knowledge points.

[@klejdisevdari3916]

38:56 Distribution of \hat{B}
54:45 MSE (Quadratic risk) of \hat{B} is the trace of the covariance matrix.

[@abdelkarimjaja7740]

thanks for these lectures

[@not_amanullah]

thanks ♥️🤍

[@chenchen2195]

A question: for MLE of the multivariate case (for example, see slides on 57:45), does the expression of quadratic risk of \hat{\beta} implies that the quadratic risk goes to zero when n becomes large (i.e., more and more data)?

[@xuchuan6401]

Yes, the risk approximates σ2*p/n when n large

[@brainstormingsharing1309]

👍👍👍👍👍

[@raffaeleiannuzzi8371]

this was hard

[@lzh00]

Agreed, I couldn't quite follow

[@not_amanullah]

this is helpful ♥️🤍