Would you like to inspect the original subtitles? These are the user uploaded subtitles that are being translated:
1
00:00:10,710 --> 00:00:14,090
Hi, everyone, and welcome in this new video package.
2
00:00:17,960 --> 00:00:25,130
I have a run and welcome in this new video in this video while going to explain theologically how the
3
00:00:25,130 --> 00:00:26,690
linear regression works.
4
00:00:27,170 --> 00:00:34,280
So to do it, I will take I have a run and welcoming this new video.
5
00:00:34,670 --> 00:00:42,710
In this video, we're all going to talk about how the linear regression works, theoretically to do
6
00:00:42,710 --> 00:00:42,950
it.
7
00:00:43,100 --> 00:00:46,310
I have just take out an extract of my book.
8
00:00:47,300 --> 00:00:47,690
So
9
00:00:50,630 --> 00:00:59,320
the first thing that you need to know is that the linear regression is the most easiest algorithm in
10
00:00:59,330 --> 00:01:00,170
machine learning.
11
00:01:01,180 --> 00:01:10,640
Indeed, to compute the error between the prediction and the real value.
12
00:01:12,770 --> 00:01:21,440
We just do the prediction minus the real value, and we put the difference at this.
13
00:01:23,660 --> 00:01:34,820
Then we do the sum of the difference for each observation, and we have the total error of the linear
14
00:01:34,820 --> 00:01:35,480
regression.
15
00:01:36,290 --> 00:01:49,040
And the goal of the linear regression is to minimize this error because it will mean that if you have
16
00:01:50,360 --> 00:01:58,430
a near equal to zero, your linear regression predict perfectly you dataset.
17
00:02:00,020 --> 00:02:07,310
So the intuition of the linear regression can be resume by and that go, even that tries to find the
18
00:02:07,310 --> 00:02:12,350
best way to minimize the distance between the predicted and the real values.
19
00:02:12,860 --> 00:02:23,960
Exactly like this in red that we have the linear regression line and each blue point is an observation,
20
00:02:24,530 --> 00:02:35,810
and the goal of the linear regression is to find a way to generalize the behavior of the blueprint.
21
00:02:37,070 --> 00:02:47,780
Here we can see that the linear regression minimize the distance between each observation.
22
00:02:48,140 --> 00:02:57,500
But in reality, this line cannot fit all the blueprint.
23
00:02:57,770 --> 00:03:06,920
The real goal is to minimize the distance between all the bluepoint, because if you find an algorithm
24
00:03:07,100 --> 00:03:12,890
in a train sets, we are going to see what is a trend set in the next video.
25
00:03:13,370 --> 00:03:24,110
But if you have some algorithm, too much train when you want to use it, for example, to do prediction
26
00:03:24,380 --> 00:03:30,500
with unknown observation, you will have very bad results.
27
00:03:31,490 --> 00:03:41,690
So really, the goal is not to have a line that fits with all the points, but it's really a line that
28
00:03:41,720 --> 00:03:47,090
generalize the behavior of each of all the points together.
29
00:03:49,550 --> 00:03:55,770
Here I have put a little resume of all you need to know about linear regression.
30
00:03:55,790 --> 00:04:05,750
So I have we need you a quick resume of this extract of my book, but I will invite you to read it because
31
00:04:05,960 --> 00:04:08,780
it is attached to this video.
3373
Can't find what you're looking for?
Get subtitles in any language from opensubtitles.com, and translate them here.