1
00:00:01,430 --> 00:00:08,330
Next, important thing we need to know is how to handle the categorical variables in a set of independent

2
00:00:08,330 --> 00:00:08,840
variables.

3
00:00:10,360 --> 00:00:18,250
If you remember, in our dataset, we have variables like airport, which has values yes and no, and

4
00:00:18,250 --> 00:00:22,580
there is a variable called waterboarding, which has values such as lake.

5
00:00:22,640 --> 00:00:23,760
There were several.

6
00:00:25,290 --> 00:00:27,740
No regression model works on real numbers.

7
00:00:28,860 --> 00:00:33,120
If you don't understand a non numerical variable such as airport.

8
00:00:35,140 --> 00:00:36,570
How can we handle this problem?

9
00:00:38,770 --> 00:00:44,770
It is simple, we just need to assign a numerical value to each of the categories of that variable.

10
00:00:46,720 --> 00:00:49,960
If the category names are numbers, the models can run.

11
00:00:51,490 --> 00:00:55,990
But we cannot assign any number to the categories as this will impact the modern.

12
00:00:57,150 --> 00:00:59,100
We need to follow proper process.

13
00:01:00,280 --> 00:01:08,200
So this process of properly creating a dummy variable, which will represent an actual variable containing

14
00:01:08,200 --> 00:01:11,710
non numerical values, is called a dummy variable creation.

15
00:01:13,830 --> 00:01:18,440
So why we do it, because regression analysis cannot handle non numeric data.

16
00:01:19,590 --> 00:01:26,250
And how we do it, we first create a new variable or set of variables, that is the dummy variable.

17
00:01:27,490 --> 00:01:33,750
And these new variables are given values of either zero or one and not anything else.

18
00:01:36,000 --> 00:01:36,600
But why?

19
00:01:37,050 --> 00:01:40,880
Why not reassign values like one, two and three, etc.?

20
00:01:44,230 --> 00:01:47,440
Let us look at the categories in this example for clarification.

21
00:01:49,220 --> 00:01:55,550
On the left, I have students and their favorite subject, on the right, I'm calling this subject.

22
00:01:56,610 --> 00:02:02,010
The first step of the military creation is creating dummy variables.

23
00:02:04,120 --> 00:02:10,960
How many variables do we need to remember this, the number of dummy variables required to replace a

24
00:02:10,960 --> 00:02:14,470
variable is number of categories in that variable, minus one.

25
00:02:15,970 --> 00:02:22,120
That is, if there are two categories we need to minus one, the military rules, that is only one variable.

26
00:02:22,750 --> 00:02:27,070
If you have three categories, like in our example, we need three minus one.

27
00:02:27,100 --> 00:02:28,520
That is to the variables.

28
00:02:30,190 --> 00:02:33,580
So I have three subjects here, science, English and maths.

29
00:02:34,930 --> 00:02:40,090
So I have created two dummy variables, one for science and one for Max.

30
00:02:41,110 --> 00:02:46,360
Science has value of one if the subject is science and it has value zero.

31
00:02:47,700 --> 00:02:51,180
So if you look at the first two, it has scientism one.

32
00:02:52,180 --> 00:02:54,840
And the other two are the subject is English and math.

33
00:02:54,940 --> 00:02:55,960
It has value zero.

34
00:02:57,690 --> 00:03:06,750
Matt has value one, if this object as bad as it has value zero, so the fourth rule as one in the match

35
00:03:06,750 --> 00:03:09,180
column and all the other rules as zero.

36
00:03:11,280 --> 00:03:20,040
This means if both are you, then this object is English, so the third row, as both you do with representing

37
00:03:20,340 --> 00:03:21,510
the third subject.

38
00:03:24,090 --> 00:03:28,070
Clearly, these zeros and ones are representing each category.

39
00:03:29,100 --> 00:03:30,570
Of the favorite classmate, Evelyn.

40
00:03:33,930 --> 00:03:40,050
Now, to answer the question why we created two dummy rebels, then we could have created a single one

41
00:03:40,170 --> 00:03:44,460
with value, such as one for science, two formats and three for English.

42
00:03:46,050 --> 00:03:49,100
This is because one, two, three have more meaning.

43
00:03:49,650 --> 00:03:50,220
That is.

44
00:03:51,920 --> 00:03:54,620
So it, too, is actually two days of fun.

45
00:03:56,010 --> 00:03:58,530
But match is not higher than our modern times.

46
00:04:00,740 --> 00:04:03,470
Math, science and English are nominal data.

47
00:04:04,160 --> 00:04:07,160
It just says the name of the favorite subject.

48
00:04:08,030 --> 00:04:11,180
So we cannot assign numerical values which can be ordered.

49
00:04:12,430 --> 00:04:17,830
Whereas when we are sending zero and one, it only represents true or false or and signal.

50
00:04:19,390 --> 00:04:22,780
Therefore, we create in minus one dummy variables.

51
00:04:24,630 --> 00:04:31,740
The value of each dummy variable is one for one single category, and the category has all.

52
00:04:33,350 --> 00:04:37,040
As zero value for all the and minus one dummy variables.

53
00:04:39,840 --> 00:04:46,110
How to interpret the results of a regression analysis containing dummy variables ordered like this will

54
00:04:46,110 --> 00:04:48,870
be discussed after we ran the aggression on our data.