WEBVTT - This file was automatically generated by VIMEO

0
00:00:07.720 --> 00:00:11.980
Hi. In this video we will learn more about the live G B M algorithm.

1
00:00:14.310 --> 00:00:15.920
Live. G B M is, uh,

2
00:00:15.920 --> 00:00:19.960
another very effective algorithm used for many modeling problems, uh,

3
00:00:19.960 --> 00:00:24.080
which is widely adopted, although a little bit less than IG Boost.

4
00:00:25.030 --> 00:00:29.200
It's an supervised learning algorithm used for classification

5
00:00:29.370 --> 00:00:34.120
regression and ranking problems, same as IG Boost. Uh,

6
00:00:34.270 --> 00:00:38.920
with XG Boost, they share a few characteristics. Uh, they are both, uh,

7
00:00:38.920 --> 00:00:43.560
gradient boosting and methods. They support missing values,

8
00:00:43.900 --> 00:00:46.480
uh, GPUs, and also parallelism.

9
00:00:48.180 --> 00:00:49.920
In case of supply. G B M,

10
00:00:50.260 --> 00:00:54.840
it directly supports categorical variables in contrary to XG Boost

11
00:00:55.050 --> 00:00:59.500
where you have to encode them first. Also internally,

12
00:01:00.120 --> 00:01:03.510
AI G B M works a little bit different. Uh,

13
00:01:03.770 --> 00:01:07.350
it uses a different three growth method, uh, than XG Boost.

14
00:01:08.510 --> 00:01:11.930
And also because of this, uh, live G B M is, uh,

15
00:01:12.450 --> 00:01:14.650
a little bit more prone to over fitting.

16
00:01:16.670 --> 00:01:18.490
In our next video, we will make a,

17
00:01:18.610 --> 00:01:22.770
a live G B M model where we will solve a multi classification problem.
