1
00:00:01,750 --> 00:00:08,350
In this session, we went to see how to handle the missing values, missing values are very much a reality.

2
00:00:09,560 --> 00:00:10,460
When we develop.

3
00:00:11,570 --> 00:00:18,530
Machine learning solutions you how to handle missing values if you want to have a higher focus, because

4
00:00:19,370 --> 00:00:25,100
sometimes the machine learning models may even throw up errors that are missing values in your data

5
00:00:25,100 --> 00:00:25,410
set.

6
00:00:25,640 --> 00:00:26,070
OK.

7
00:00:26,360 --> 00:00:30,810
And more fundamentally, missing values can lead to a biased model.

8
00:00:31,680 --> 00:00:35,290
You can make a wrong classification to the extent of missing values.

9
00:00:35,300 --> 00:00:38,060
It's more like why?

10
00:00:38,120 --> 00:00:41,330
Why do we have missing values in the first place?

11
00:00:41,930 --> 00:00:48,920
Can be because values can be lost when you're extracting data from from some of the system.

12
00:00:49,790 --> 00:00:54,380
Right are at the time of data collection, some values might not be collected.

13
00:00:54,380 --> 00:00:55,750
It can be due to an error.

14
00:00:55,760 --> 00:00:58,520
Also, there can be various reasons.

15
00:00:58,850 --> 00:01:01,100
But it's a it's a reality.

16
00:01:01,100 --> 00:01:02,680
It's something that you cannot avoid.

17
00:01:03,080 --> 00:01:03,400
Right.

18
00:01:03,440 --> 00:01:06,790
You will have some missing values.

19
00:01:07,790 --> 00:01:08,050
Right.

20
00:01:08,330 --> 00:01:14,330
So how do we handle it, first strategy, then second strategy.

21
00:01:15,760 --> 00:01:25,750
Use the value of mean or median, if it is up, if that particular missing value should have a continuous

22
00:01:25,750 --> 00:01:33,340
data, like if the missing value corresponds to a categorical data use, more like you've already seen

23
00:01:33,340 --> 00:01:36,190
that what is mean median mode.

24
00:01:37,630 --> 00:01:39,730
Let's call DC and come back.

25
00:01:39,730 --> 00:01:40,130
Right.

26
00:01:40,550 --> 00:01:46,330
If you see this, this is something we saw in the previous session molders that frequently occurring

27
00:01:46,330 --> 00:01:52,090
number and median is the midpoint after arranging data in ascending order.

28
00:01:52,750 --> 00:01:54,670
Mean is the arithmetic average.

29
00:01:55,630 --> 00:01:55,880
Right.

30
00:01:56,240 --> 00:02:07,150
So if you're how missing values corresponding to a continuous data you use either mean or maybe.

31
00:02:08,170 --> 00:02:15,610
If the missing data corresponds to a categorical type of data, it was more.

32
00:02:17,400 --> 00:02:19,580
That's all very simple, right?

33
00:02:20,150 --> 00:02:30,440
So in our data, in the insurance case study that we are looking at, we have 149 nonviolence and even

34
00:02:30,440 --> 00:02:37,320
taking nonviolence according to different factors like different variables.

35
00:02:38,480 --> 00:02:43,730
Now, why I'm taking it, because my strategy will be different for different types of variables.

36
00:02:44,090 --> 00:02:50,030
If the variable is continuous, I replace with minimal medium in the variable is categorical.

37
00:02:50,030 --> 00:02:51,050
I replace it more.

38
00:02:51,320 --> 00:02:51,530
Right.

39
00:02:51,650 --> 00:02:54,980
Which is the highest frequency value.

40
00:02:55,820 --> 00:02:56,140
Right.

41
00:02:56,840 --> 00:02:59,300
So now I am replacing.

42
00:02:59,300 --> 00:03:00,040
I have no water.

43
00:03:00,590 --> 00:03:09,380
I am replacing with me in the case of a continuous variable, in the case of a categorical variable

44
00:03:09,380 --> 00:03:12,290
and replacing with more, as you can see the code here.

45
00:03:13,450 --> 00:03:19,630
And that on this takes care of the missing girls because you are replacing with the appropriate values.

46
00:03:19,630 --> 00:03:24,100
So now your dataset is complete and you're ready to move forward, right.

47
00:03:24,600 --> 00:03:29,530
Very, very simple technique, but a very powerful technique in handling missing girls.

48
00:03:29,980 --> 00:03:38,020
As I said, missing values must be addressed appropriately, either delete or replaced with mean million

49
00:03:38,020 --> 00:03:38,460
or more.

50
00:03:39,160 --> 00:03:39,500
Right.

51
00:03:39,970 --> 00:03:43,630
So this completes this short session.

52
00:03:43,870 --> 00:03:52,300
OK, so we go on to the next Silicon Valley, be talking about atrazine polglase, OK.

53
00:03:53,710 --> 00:04:01,990
Just one last point, after replacing Chip once, if there are any missing letters.

54
00:04:03,120 --> 00:04:03,420
Right.

55
00:04:04,030 --> 00:04:09,990
Please ensure there are no missing just to do the check, right, to be absolutely sure that there are

56
00:04:09,990 --> 00:04:11,230
no missing ones.

57
00:04:11,430 --> 00:04:16,470
I found the code again and see that I need the same values as you can see in the case study that we

58
00:04:16,470 --> 00:04:20,370
have taken from the caller in on the missing values.

59
00:04:20,460 --> 00:04:23,460
And we're good to Moyet, right.

60
00:04:24,360 --> 00:04:24,720
OK.