1
00:00:01,530 --> 00:00:03,950
This session is about variable transmission.

2
00:00:04,800 --> 00:00:10,190
We are trying to derive maximum value out of what our data got.

3
00:00:11,410 --> 00:00:12,910
Incredible transformation.

4
00:00:13,210 --> 00:00:21,130
You're not having any data like you're only trying to make the data more useful and different techniques

5
00:00:21,170 --> 00:00:23,890
are available for variable transmission.

6
00:00:25,250 --> 00:00:32,060
That is transformation of continuous variable, and that is transformation of categorical, like both

7
00:00:32,060 --> 00:00:39,150
we will see in this session in the case of continuous variable, which is numerically what's right.

8
00:00:39,170 --> 00:00:48,710
You can transform the data by converting the data from a numeric to a laboratory or square root on human

9
00:00:49,580 --> 00:00:56,720
right through this, we're essentially changing the nature of relationship on nature, of distribution,

10
00:00:56,720 --> 00:00:57,470
of dataset.

11
00:00:58,280 --> 00:01:00,710
That's what we are doing through this exercise.

12
00:01:01,400 --> 00:01:11,780
Beny is transforming a new data into a category like we already saw the example of beaning and logarithmic

13
00:01:12,590 --> 00:01:14,720
in treating outlier data.

14
00:01:15,380 --> 00:01:16,870
So that's something we Soldini.

15
00:01:17,930 --> 00:01:26,030
In the case of categorical data, the transformation to numeric value happens through the process of

16
00:01:26,030 --> 00:01:27,790
what is known as encoding, OK?

17
00:01:28,220 --> 00:01:35,210
In fact, the transformation is more needed in categorical variables than numeric variables.

18
00:01:35,870 --> 00:01:42,450
This is because categorical variables as such cannot be used directly in most machine learning models.

19
00:01:42,470 --> 00:01:49,910
Hence, it becomes a necessity for us to convert the categorical variables into numeric variables.

20
00:01:50,870 --> 00:01:51,170
Right.

21
00:01:51,980 --> 00:02:01,460
So if you see the insurance example, the ones that are mentioned in green right there on non numeric

22
00:02:01,880 --> 00:02:08,840
categorical, the ones that are mentioned in red, they are continuous numeric variables.

23
00:02:09,380 --> 00:02:16,970
So all the ones that are that are shown in red, the numeric variables, I can translate them if needed

24
00:02:16,970 --> 00:02:20,650
into longer to make a square root or Kubelik.

25
00:02:21,170 --> 00:02:21,530
Right.

26
00:02:21,770 --> 00:02:30,440
Whereas the ones in green non numeric I have to mandatorily convert them, transform them into numeric

27
00:02:30,440 --> 00:02:35,690
data, only then I can use them for machine learning models.

28
00:02:36,200 --> 00:02:36,550
Right.

29
00:02:36,860 --> 00:02:41,200
So in the case of non numeric or categorical, it's more than necessary for.

30
00:02:42,470 --> 00:02:42,820
Right.

31
00:02:44,360 --> 00:02:48,260
So how do you convert Python comes with preloaded library?

32
00:02:49,040 --> 00:02:53,880
As you can see the example here, I have six smoker region, right?

33
00:02:53,900 --> 00:02:58,570
These are categorical variables using the encoding process.

34
00:02:58,690 --> 00:03:06,170
I input pre processing and then I use pre dot, label encoder and then I transform and you can see that

35
00:03:06,380 --> 00:03:15,710
the categorical data is now converted into numeric data and this data can now be used for generating

36
00:03:15,710 --> 00:03:16,400
machine learning.

37
00:03:16,400 --> 00:03:19,610
Models get very simple, right?

38
00:03:20,120 --> 00:03:23,420
So variable transformation is a very important process.

39
00:03:24,320 --> 00:03:31,730
More important in the case of categorical variables, because I cannot use categorical variables directly

40
00:03:31,730 --> 00:03:37,700
in machine learning models like they must be first converted into numeric value.

41
00:03:38,790 --> 00:03:46,180
OK, so that completes the session on variable transmission, right?

42
00:03:46,290 --> 00:03:53,100
In fact, with that session we completed and are exploratory data analysis online program.

43
00:03:53,550 --> 00:03:58,660
So just to summarize, we started with understanding the data, basic understanding of the data.

44
00:03:59,340 --> 00:04:07,680
Then we did univariate analysis, then Vivarium analysis to see the relationship between two variables.

45
00:04:08,040 --> 00:04:13,240
We also saw the concept of Collini are clearly something we don't want in our dataset.

46
00:04:13,890 --> 00:04:20,370
Then we also saw missing values, treatment, how to handle missing values, how to handle outliers,

47
00:04:20,730 --> 00:04:27,980
how to transform variables like andas we saw as part of exploratory data analysis.

48
00:04:28,380 --> 00:04:28,710
Right.

49
00:04:29,070 --> 00:04:36,930
So if you really see the last three, what I would call a future Regini because I'm trying to get maximum

50
00:04:36,930 --> 00:04:44,460
value out of whatever data and I want to make the data more useful for my purposes.

51
00:04:45,700 --> 00:04:46,040
Right.

52
00:04:46,810 --> 00:04:52,990
So that completes the session on exploratory data analysis.

53
00:04:53,830 --> 00:04:54,320
Thank you.

54
00:04:54,610 --> 00:04:56,160
And I wish you the best of luck.