1
00:00:00,920 --> 00:00:05,480
In this video, we will learn how to handle the outlying values in our dataset.

2
00:00:06,560 --> 00:00:15,560
We have identified from univariate analysis that in hot rooms contained exceptionally large values.

3
00:00:17,280 --> 00:00:20,610
These values are outliers and we have to treat them.

4
00:00:22,640 --> 00:00:29,560
From the lecture, we have told you that when we have an outlier, we cap the values in the variable

5
00:00:29,990 --> 00:00:37,490
to upper limit to find the upper limit, we find either a non outlaying value and find a multiplication

6
00:00:37,490 --> 00:00:37,930
factor.

7
00:00:39,390 --> 00:00:44,500
All we use the 99 percent daily value and use the multiplication factor to get the upper value.

8
00:00:45,420 --> 00:00:49,680
We don't change the values of all the old values to this upper value.

9
00:00:52,480 --> 00:01:00,550
To do the same thing in and courtrooms, we need to first identify those cells which contain the outline

10
00:01:00,560 --> 00:01:01,030
values.

11
00:01:02,530 --> 00:01:11,800
To identify the cells, we are going to apply a filter first to apply filters on the table, we select

12
00:01:11,800 --> 00:01:16,510
the top row of our table, then we go to data.

13
00:01:18,020 --> 00:01:20,000
And we click on this filter option.

14
00:01:23,770 --> 00:01:30,210
You can see that these additional small boxes containing a small triangle are coming with each column.

15
00:01:31,660 --> 00:01:37,940
These boxes are giving us the filtering and sorting option for this column here.

16
00:01:37,990 --> 00:01:44,500
You can see that all the values of this column are listed in this small box.

17
00:01:45,800 --> 00:01:50,970
So if we go to any rooms now and open this small box for.

18
00:01:53,690 --> 00:02:00,410
You can see the values are starting from ten point zero five and it goes on and suddenly the values

19
00:02:00,410 --> 00:02:06,320
increase from fifteen point four to eighty one point one two and then one hundred one point one two.

20
00:02:07,790 --> 00:02:11,570
So we have to treat these two exceptionally large values.

21
00:02:13,590 --> 00:02:21,270
And to get the valley of upper limit, we are going to use this largest normal value.

22
00:02:23,190 --> 00:02:27,300
Which is fifteen point four, and we will use a multiplication factor of three.

23
00:02:28,140 --> 00:02:34,140
So the value we are going to replace these outlying values is three and to fifteen point four, which

24
00:02:34,140 --> 00:02:35,190
is forty six point two.

25
00:02:37,270 --> 00:02:45,310
So to select these two cells will first and check all of them and go and check these last three cells.

26
00:02:49,820 --> 00:02:54,950
You can see we have only those observations with us, which have these three types of values.

27
00:02:56,260 --> 00:03:03,340
Now, to change the value in this particular cell, to train to fifteen point four, we will rate is

28
00:03:03,340 --> 00:03:08,320
equal to three in do fifteen point four.

29
00:03:12,930 --> 00:03:16,230
For this also, we will continue to train to fifteen point four.

30
00:03:21,220 --> 00:03:28,540
So this method of treating outliers, that is we create an upper limit and change the outlying values

31
00:03:28,540 --> 00:03:30,710
to that upper limit is very subjective.

32
00:03:31,360 --> 00:03:35,560
You can choose any non Nordling value.

33
00:03:35,590 --> 00:03:42,280
You can choose any multiplication factor to establish the value and change these outlying values according

34
00:03:42,280 --> 00:03:43,150
to your business need.

35
00:03:44,290 --> 00:03:49,270
So with these actions, we have treated the outlying values in our dataset.

36
00:03:50,110 --> 00:03:54,280
Now we will take this select Altec box to get the whole table.

37
00:03:55,450 --> 00:03:59,920
Now you can see that we do not have those two exceptionally large values.

38
00:04:00,340 --> 00:04:01,440
We have this value.

39
00:04:01,990 --> 00:04:05,930
If you think this is still too big, you can treat it further.

40
00:04:05,980 --> 00:04:07,640
You can get a smaller value here.

41
00:04:08,710 --> 00:04:14,530
For now, we are going to keep this value as the upper limit porras and courtroom's variable.