Topic 10. Part 1. Gradient boosting basics

[@rrrprogram8667]

Subscribed.... Nice video... Love to see more

[@consumidorbrasileiro222]

hi, yury. thanks for the demo. could you give any explanation for why linear classifier works better for high dimensional data?

[@festline]

hi, sorry for a late reply. that's the whole topic #5 where we discuss Forests as opposed to linear models. Basically, these are trees that do not work well in case of high dimensional data (also, KNN and other metric-based algos, but tree-based methods are by far the most important after linear models). So linear models might work better than tree-based methods in case of many features.

[@AbsoluteArtist0709]

Hello Yury, thank you for the excellent lecture! Is the gradient boosting algorithm implementation assignment still available? I can't seem to find it...only predicting flight delays with XGBoost. Thank you!

[@festline]

Hi! It's a good assignment, and I plan to release full version of assignments later this year for my Patreon supporters.

[@Shakarim94]

Great lecture, just a small correction: it should be the squared L2 norm

[@festline]

sorry, can you point out the exact time stamp?

[@festline]

In 38:20 it indeed shall be squared L2 norm. Thanks!

[@pavanilla4374]

sir start course on natural language processing

[@festline]

check out cs224n

[@vnnw]

No HD video available?

[@festline]

If youtube doesn't show, then no