KD-Trees and Range search

[@yusufahmed2233]

I love how you are actually drawing instead of animating it, and the quality is not less in any way! Thank for the informative and fun video

[@alkodjdjd]

Finally a video without math crap and straight to the point. Congrats!

[@uttarandey405]

One of the best explanation I've ever seen in my life....thanks a lot...

[@spandankumarsahu3313]

Excellent video :) Saved so many pages of non-enticing text :)

[@Lukas-wm8dy]

This video should have much more views, really well explained. Thanks a lot!

[@linuxguru27]

I liked your explanation, Kudos! Thank you

[@ama_do_it]

My left ear said this lecture was fantastic

[@irajulsaxena]

Thank You so much for the knowledge.... also the effort to make such a great, well explained video.... and editing it too... all too good 😊👍

[@bikers11254]

+1

[@tomwimmenhove4652]

Well done, great explanation. Thanks!

[@Vivekanand-q7c]

Thanks for this nice explanation

[@hobobmo]

Excellent explanation. Thanks!

[@osolovyoff]

Thank you so much !!! I fully understood it after your video

[@kristiant.srensen9400]

Great explanation, thanks a lot!

[@edvogel56]

Also a great way to make Mondrian like paintings

[@martinliptak4781]

Thank you, funny, informative and helpuf!

[@user-gm3gt7cl2y]

cool explanation, thanks

[@MaheshKumar-yz7ns]

Very Good Explanation.

[@filipf7983]

Thanks a lot!

[@mohitkatyal7988]

Great Video, thanks so much!

[@amankhurana2154]

Awesome, Perfect :)

[@tahmidziko2601]

Superb just like an awesome

[@mikl2345]

I tried to implement kd trees without looking at the solution first. I came up with a node with four links, one for each possible pair of positive/negative direction in each dimension (i.e., each of these four links denotes a quadrant). Then I looked at the solution and was surprised that it seemed more complex, because it alternates x and y directions rather than choosing a quadrant in a uniform way at each step, as my 'solution' did. But now I think my solution seems flawed. Is it possible from what I've said to explain why? Putting it another way, why does the kd tree as shown here need to alternate between dimensions? Thanks for a great video!

[@niklaslundstrom1235]

So it seems like what you are describing is a KDB-Tree, which is a perfectly fine data structure. KD-Trees are a generalisation of binary trees in higher dimensions, whereas KDB-Trees are a generalisation of b-trees in higher dimension, i.e. the possibility for a node to have more than two children. I think it can be a good idea to think about how you want to split at each level. By for example splitting over x and y independently _____x______ | | | y|-----|--------| y |___|_____| x as you did can be fine for some purposes, but could also lead to a more unbalanced tree. Imagine for example we have a bunch of points in the upper-left corner and lower-right corner, but very few in lower-left and upper-right corner. If we want a balanced tree, we would have to split x and y dependent on each other _____x________ y|-----| | | |------------| y |___|________| x and thus first checking x, and then y. This would of course be possible with a KDB-Tree as well but it would still be necessary to check one dimension at the time, but then output four children.

[@xiaoyujia9637]

coooool ！ easy to understand，thanks a lot

[@cq6269]

Thank you

[@HebrewSongPod]

well explained!

[@dhruv6433]

welcome IITD peeps COL106😁

[@royleon5]

thought someone was talking behind me..

[@questforprogramming]

Explanation is good, but bgm ruined my concentration....please keep doing videos but without bgm. Thank you

[@niklaslundstrom1235]

all right, thanks for the input :)

[@soumyajitdutta428]

At 2:57-3:03 I hear a siren, Am i right?

[@Darieee]

nah my phone was just ringing

[@billwang1456]

2:52

[@doekewartena5729]

This is a terrible way to do a range search on a k-d tree. You kind of want to define a branch that goes left-ish and one that goes right-is. And then add the branches inbetween those two without checking the individual points. Sorry for my poor explanation.