1
00:00:01,780 --> 00:00:08,930
To handle the outliers in any hard rooms and rainfall, we will use the capping and flooring technique

2
00:00:09,170 --> 00:00:10,130
discussed in Lichter's.

3
00:00:11,410 --> 00:00:18,670
We will cap the upper value of and hot rooms to a value of three into 99 percentile, and we will cap

4
00:00:18,670 --> 00:00:22,650
the lower value of for two point three into first person date.

5
00:00:24,810 --> 00:00:30,010
For this, we need to know how to find out the value of a variable at a particular percentile level.

6
00:00:31,360 --> 00:00:38,770
The method for this is to use quintile to relate quintile within bracket.

7
00:00:39,970 --> 00:00:40,310
Relate.

8
00:00:40,570 --> 00:00:42,700
D.F. dollar and hard rooms.

9
00:00:47,290 --> 00:00:51,600
Coma, we will mention the percentile values or zero point nine.

10
00:00:52,570 --> 00:00:53,320
If you'd unders.

11
00:00:55,460 --> 00:01:04,880
In the desert, you can see fifteen point three nine nine is the 99 percentile value in in hot rooms

12
00:01:06,190 --> 00:01:10,310
to be won three times of this value as the upper limit.

13
00:01:10,460 --> 00:01:16,010
So we will assign you V a value of three times this.

14
00:01:16,250 --> 00:01:17,810
So I'll copy paste.

15
00:01:20,680 --> 00:01:21,330
This lane.

16
00:01:25,400 --> 00:01:33,290
So about well, you get a number, which is forty six point two, which is three times fifteen point

17
00:01:33,290 --> 00:01:33,800
three nine.

18
00:01:35,200 --> 00:01:40,530
Now, for all we're losing in hot rooms, we will compare whether it is greater than you'll.

19
00:01:41,380 --> 00:01:44,250
And if it is, we will see indeed that value to you.

20
00:01:44,260 --> 00:01:46,030
We to do that.

21
00:01:46,630 --> 00:01:47,250
We will wait.

22
00:01:48,540 --> 00:01:50,640
Beef, dollar and hard rooms.

23
00:01:54,010 --> 00:01:57,630
Within square brackets, we relate B F dollar and hard rooms.

24
00:01:58,970 --> 00:02:00,200
Greater than EUI.

25
00:02:07,260 --> 00:02:14,330
So all values of this variable greater than you, we should get the value you we.

26
00:02:17,460 --> 00:02:25,710
Now, if you're on this, all the values which were beyond the movie range now have value, U.B..

27
00:02:26,980 --> 00:02:34,810
Now, if we run UDD on this particular variable, we will somebody and within bracket will ADF dollar

28
00:02:35,770 --> 00:02:36,460
and hotrods.

29
00:02:39,110 --> 00:02:39,680
Run this.

30
00:02:41,150 --> 00:02:45,950
You can see the median and mean are now much closer to.

31
00:02:47,740 --> 00:02:52,490
And the maximum value is change to 46, which was earlier hundred and one.

32
00:02:54,390 --> 00:02:56,560
So let's let us do the same thing for rainfall.

33
00:02:56,860 --> 00:03:04,060
Rainfall has outlaid on the lower side, so we assign Elvie the value of point three times the first

34
00:03:04,060 --> 00:03:13,390
quartile value to relate Elway's equal to zero point three times gone by.

35
00:03:20,060 --> 00:03:22,730
And the variable is D.F. dollar rainfall.

36
00:03:24,200 --> 00:03:25,790
And we want DeForest COINTEL.

37
00:03:26,000 --> 00:03:27,290
So at this point due to one.

38
00:03:29,690 --> 00:03:30,620
Do we have the law?

39
00:03:30,680 --> 00:03:32,440
Well, you know, we've been done this.

40
00:03:33,340 --> 00:03:37,960
There is another variable, Elvie, here, which has a value of six.

41
00:03:40,030 --> 00:03:41,790
No, let us replace this value.

42
00:03:42,450 --> 00:03:47,970
We will ride the F dollar rainfall.

43
00:03:48,690 --> 00:03:55,340
And within square brackets, we will light D.F. dollar rainfall is less than this first court quantize

44
00:03:55,350 --> 00:03:55,830
value.

45
00:03:57,450 --> 00:04:01,980
These values will get Elvie input on this.

46
00:04:03,330 --> 00:04:04,680
The values have been changed.

47
00:04:04,860 --> 00:04:11,390
If we done EGD again on rainfall by somebody, B.F. dollar rainfall.

48
00:04:17,040 --> 00:04:23,250
We again see that mean and median values are quite close now, and the minimum value, which was earlier,

49
00:04:23,640 --> 00:04:25,980
probably three, is now six.

50
00:04:26,550 --> 00:04:32,930
If you want to have a value of minimum value closer to the first quintile, you we could have used a

51
00:04:32,940 --> 00:04:35,760
different multiplier than zero point three.

52
00:04:36,690 --> 00:04:38,490
Probably that gave us a very low value.

53
00:04:39,360 --> 00:04:43,300
But still, the median MediaNet quite close and we are happy with this.

54
00:04:43,690 --> 00:04:48,750
So we have done the outlash treatment for the two variables that we wanted to treat.