1
00:00:00,200 --> 00:00:02,600
Hello and welcome to this art tutorial.

2
00:00:02,600 --> 00:00:04,600
Well, I. Hope the homework went well.

3
00:00:04,600 --> 00:00:06,133
I hope you've.

4
00:00:06,133 --> 00:00:09,900
Obtained some interesting results
and without waiting,

5
00:00:09,900 --> 00:00:13,800
I'm going to complete the backward
elimination here on R.

6
00:00:13,900 --> 00:00:17,300
So that you can see if you obtain
the same results as what we are about to.

7
00:00:17,300 --> 00:00:20,100
Obtain, and mostly
so that you can see how we complete.

8
00:00:20,100 --> 00:00:20,633
Backward.

9
00:00:20,633 --> 00:00:23,200
Elimination and are up. To the end.

10
00:00:23,200 --> 00:00:24,100
So let's do that.

11
00:00:24,100 --> 00:00:26,433
So in the previous tutorial we. Went up.

12
00:00:26,433 --> 00:00:30,500
To building this regressor
without the state.

13
00:00:30,500 --> 00:00:32,233
Independent variable.

14
00:00:32,233 --> 00:00:33,466
And that's completed the.

15
00:00:33,466 --> 00:00:34,800
Step five here.

16
00:00:34,800 --> 00:00:36,600
Of the backward elimination algorithm.

17
00:00:36,600 --> 00:00:37,900
And so now we need to.

18
00:00:37,900 --> 00:00:40,333
Go back to step three to look.

19
00:00:40,333 --> 00:00:43,300
Again for the independent variable
that. Has the highest p value.

20
00:00:43,300 --> 00:00:46,300
And then we need to compare it
to the significance level.

21
00:00:46,400 --> 00:00:47,133
To decide.

22
00:00:47,133 --> 00:00:49,533
If we need to remove this
independent variable.

23
00:00:49,533 --> 00:00:51,000
With the highest p value.

24
00:00:51,000 --> 00:00:52,000
So let's do this.

25
00:00:52,000 --> 00:00:54,033
It's actually already ready.

26
00:00:54,033 --> 00:00:56,466
We just need to select this.

27
00:00:56,466 --> 00:00:57,933
And press Command and control.

28
00:00:57,933 --> 00:01:00,000
Press enter to execute.

29
00:01:00,000 --> 00:01:02,033
And here it is perfect.

30
00:01:02,033 --> 00:01:03,866
So let's move that up.

31
00:01:04,866 --> 00:01:05,733
All right.

32
00:01:05,733 --> 00:01:08,033
So as you can see. Here
we have the new formula.

33
00:01:08,033 --> 00:01:09,166
Prophet expressed.

34
00:01:09,166 --> 00:01:12,233
As a linear combination
of all the independent variables.

35
00:01:12,233 --> 00:01:14,366
Except for this state.

36
00:01:14,366 --> 00:01:15,100
So that's fine.

37
00:01:15,100 --> 00:01:16,100
And here therefore.

38
00:01:16,100 --> 00:01:18,100
We have the p values for these new.

39
00:01:18,100 --> 00:01:21,100
Team of three independent variables.

40
00:01:21,600 --> 00:01:22,000
All right.

41
00:01:22,000 --> 00:01:23,433
And now what. Do we see here.

42
00:01:23,433 --> 00:01:23,866
Okay.

43
00:01:23,866 --> 00:01:27,133
So we can see that
of course the already spent is still

44
00:01:27,300 --> 00:01:30,633
highly statistically significant
with three stars.

45
00:01:30,633 --> 00:01:32,900
Here and a very. Low p value.

46
00:01:32,900 --> 00:01:34,733
And mostly what do we need to do now.

47
00:01:34,733 --> 00:01:36,700
We need to look for the highest p value.

48
00:01:36,700 --> 00:01:39,666
And that's actually this 10.6.

49
00:01:39,666 --> 00:01:41,933
Oh to 60%.

50
00:01:41,933 --> 00:01:45,533
60% is definitely a very high
p value, way.

51
00:01:45,533 --> 00:01:47,900
Above the significance level of 5%.

52
00:01:47,900 --> 00:01:50,466
So definitely administration is not.

53
00:01:50,466 --> 00:01:52,200
Statistically. Significant.

54
00:01:52,200 --> 00:01:56,100
That is administration has no effect
on the dependent variable profit.

55
00:01:56,700 --> 00:01:58,633
So great. That's pretty clear.

56
00:01:58,633 --> 00:02:02,933
We need to remove administration
from or regression equation.

57
00:02:03,600 --> 00:02:04,766
So let's do this.

58
00:02:04,766 --> 00:02:06,233
As usual. We're going to. Copy this

59
00:02:07,400 --> 00:02:10,300
copy paste.

60
00:02:10,300 --> 00:02:13,300
And here
will simply remove administration.

61
00:02:14,400 --> 00:02:15,433
There we go.

62
00:02:15,433 --> 00:02:17,200
And now we have our new regressor.

63
00:02:17,200 --> 00:02:18,100
Ready with.

64
00:02:18,100 --> 00:02:19,933
Only two independent variables

65
00:02:19,933 --> 00:02:23,133
composed of already spent
which we already know is highly.

66
00:02:23,133 --> 00:02:24,800
Statistic. Significant.

67
00:02:24,800 --> 00:02:27,700
And the marketing spend marketing spend.

68
00:02:27,700 --> 00:02:29,700
So far the p value. Is 10%.

69
00:02:29,700 --> 00:02:31,900
Let's see what it will become.

70
00:02:31,900 --> 00:02:36,900
So I'm actually going
to build this new model by executing this.

71
00:02:36,900 --> 00:02:39,633
So command and control
plus enter to execute.

72
00:02:39,633 --> 00:02:40,533
Here we go.

73
00:02:40,533 --> 00:02:43,166
And now let's find out about.

74
00:02:43,166 --> 00:02:45,333
The statistical results of this new.

75
00:02:45,333 --> 00:02:46,833
Regression model.

76
00:02:46,833 --> 00:02:47,866
Let's select this.

77
00:02:47,866 --> 00:02:50,533
Press Command and Control
plus enter to. Execute.

78
00:02:50,533 --> 00:02:52,800
And here are the new. Statistical results.

79
00:02:52,800 --> 00:02:54,466
With only the p values

80
00:02:54,466 --> 00:02:57,733
of the two independent variables
already spent and marketing spend.

81
00:02:58,366 --> 00:02:59,200
Okay, so let's see.

82
00:02:59,200 --> 00:03:01,700
Wow there is something interesting here.

83
00:03:01,700 --> 00:03:02,733
Can you see it.

84
00:03:02,733 --> 00:03:06,300
Well first participant is still highly
statistically significant.

85
00:03:06,500 --> 00:03:08,033
No surprise about that.

86
00:03:08,033 --> 00:03:11,033
But we actually. Have one surprise here.

87
00:03:11,033 --> 00:03:15,200
Remember the p value of the marketing
spend was 10% at the previous step.

88
00:03:15,566 --> 00:03:17,366
Now it's 6%.

89
00:03:17,366 --> 00:03:19,466
And actually we want a dot here.

90
00:03:19,466 --> 00:03:20,266
You know we.

91
00:03:20,266 --> 00:03:23,333
Jumped from this category
between 10% and one.

92
00:03:23,800 --> 00:03:26,400
To this. Category between 5% and 10%.

93
00:03:26,400 --> 00:03:27,600
So we actually have.

94
00:03:27,600 --> 00:03:30,566
A dot now for this
marketing spend independent variable.

95
00:03:30,566 --> 00:03:33,600
Plus it's
very close to the significance level.

96
00:03:33,900 --> 00:03:35,633
It's very close to 5%.

97
00:03:35,633 --> 00:03:38,200
You know for example,
if our significance level was.

98
00:03:38,200 --> 00:03:40,733
7% we would have kept this.

99
00:03:40,733 --> 00:03:42,633
Independent variable marketing spend.

100
00:03:42,633 --> 00:03:44,600
So it's kind of arbitrary.

101
00:03:44,600 --> 00:03:45,733
You know, we we.

102
00:03:45,733 --> 00:03:47,700
Don't know if we really need to remove.

103
00:03:47,700 --> 00:03:50,166
This just because. Our backward
elimination.

104
00:03:50,166 --> 00:03:52,266
Tells us to, due to the fact that.

105
00:03:52,266 --> 00:03:54,166
We made an arbitrary. Choice of.

106
00:03:54,166 --> 00:03:56,700
5% for the significance level.

107
00:03:56,700 --> 00:03:58,433
So since this is a tutorial.

108
00:03:58,433 --> 00:04:01,700
About backward elimination,
we will actually remove. It.

109
00:04:01,866 --> 00:04:03,900
But this will not be our final.

110
00:04:03,900 --> 00:04:06,900
Word because at the end of this part,
there will be this.

111
00:04:06,900 --> 00:04:09,733
Section
about evaluating model's performance.

112
00:04:09,733 --> 00:04:13,033
And we will actually add a criterion
to make.

113
00:04:13,033 --> 00:04:15,066
A better call at deciding.

114
00:04:15,066 --> 00:04:18,300
If we really need to remove this
marketing spend.

115
00:04:18,333 --> 00:04:19,766
Because right now, the call

116
00:04:19,766 --> 00:04:23,566
to remove this marketing spend
independent variable is arbitrary.

117
00:04:24,100 --> 00:04:28,800
So we will remove it
because we want to follow thoroughly.

118
00:04:28,800 --> 00:04:30,866
The backward elimination algorithm.

119
00:04:30,866 --> 00:04:32,300
But keep that in mind

120
00:04:32,300 --> 00:04:35,800
that it's not a final word about deciding
what's the final time is going.

121
00:04:35,800 --> 00:04:37,766
To be, and then we will get back to this.

122
00:04:37,766 --> 00:04:40,200
Problem later in this. Course that is.

123
00:04:40,200 --> 00:04:41,300
Later in this part.

124
00:04:41,300 --> 00:04:45,433
To make a better call about whether we
need to remove yes or no marketing spent.

125
00:04:45,433 --> 00:04:47,266
So let's remove it for now.

126
00:04:47,266 --> 00:04:49,266
That's actually the final solution.

127
00:04:49,266 --> 00:04:51,500
But it's really a good choice of yours.

128
00:04:51,500 --> 00:04:56,166
If you actually decided to keep marketing
spend, you will be proud of this choice.

129
00:04:56,166 --> 00:04:57,600
At the end of this part.

130
00:04:57,600 --> 00:04:58,866
So congratulations!

131
00:04:58,866 --> 00:05:01,000
Anyway, if you decided to keep marketing.

132
00:05:01,000 --> 00:05:03,666
Spend and of course,
also congratulations to the.

133
00:05:03,666 --> 00:05:04,833
Others who decided.

134
00:05:04,833 --> 00:05:06,733
To remove marketing spend
because that means.

135
00:05:06,733 --> 00:05:08,700
You just followed thoroughly the.

136
00:05:08,700 --> 00:05:10,700
Backward elimination algorithm.

137
00:05:10,700 --> 00:05:13,066
So congratulations to both of you.

138
00:05:13,066 --> 00:05:17,233
And let's finish this backward
elimination algorithm.

139
00:05:17,233 --> 00:05:20,233
Because actually, you know,
this is definitely the final step.

140
00:05:20,566 --> 00:05:24,100
We have only one independent
variable left R&D spend.

141
00:05:24,100 --> 00:05:27,033
And we already know it's highly.
Statistically. Significant.

142
00:05:27,033 --> 00:05:28,766
But let's actually make this model.

143
00:05:28,766 --> 00:05:31,633
By selecting. And executing this.

144
00:05:31,633 --> 00:05:33,633
Here we go model created.

145
00:05:33,633 --> 00:05:35,166
And now let's have a final.

146
00:05:35,166 --> 00:05:38,100
Look at the statistical.

147
00:05:38,100 --> 00:05:39,633
Informations of our model.

148
00:05:39,633 --> 00:05:40,500
So let's press.

149
00:05:40,500 --> 00:05:42,600
Command and Control
plus enter to. Execute.

150
00:05:42,600 --> 00:05:47,033
And here are the final informations
of the final optimal team

151
00:05:47,400 --> 00:05:50,900
a team that is actually composed
of one independent variable.

152
00:05:50,900 --> 00:05:53,366
So that's actually funny
to call. It a team.

153
00:05:53,366 --> 00:05:55,500
But anyway that's just a team of.

154
00:05:55,500 --> 00:05:58,366
One that happens. But wait for the final.

155
00:05:58,366 --> 00:06:00,066
Part to see if a team.

156
00:06:00,066 --> 00:06:02,933
Of one can really be the best team.

157
00:06:02,933 --> 00:06:03,866
So that's it.

158
00:06:03,866 --> 00:06:06,266
We actually completed backward
elimination.

159
00:06:06,266 --> 00:06:10,866
Our final model is composed of only one
independent variable, the R&D spend.

160
00:06:11,366 --> 00:06:12,166
And we can clearly.

161
00:06:12,166 --> 00:06:13,966
See that this. Independent variable.

162
00:06:13,966 --> 00:06:15,266
Is highly statistically.

163
00:06:15,266 --> 00:06:17,000
Significant, not only because.

164
00:06:17,000 --> 00:06:18,433
We have three stars here,

165
00:06:18,433 --> 00:06:21,133
but also because the p value is really,
really small.

166
00:06:21,133 --> 00:06:22,500
So definitely that's a very.

167
00:06:22,500 --> 00:06:24,733
Precious information for the investors.

168
00:06:24,733 --> 00:06:26,533
That should really look at.

169
00:06:26,533 --> 00:06:28,600
This independent. Variable to add.

170
00:06:28,600 --> 00:06:31,600
Another criterion
into their investment decisions.

171
00:06:32,366 --> 00:06:33,900
So again, congratulations.

172
00:06:33,900 --> 00:06:35,366
To both of you who.

173
00:06:35,366 --> 00:06:39,233
Either found a final team
of only one independent variable,

174
00:06:39,600 --> 00:06:42,300
the R&D spend or a final team of two

175
00:06:42,300 --> 00:06:45,433
independent variables
the R&D spend and the marketing spend.

176
00:06:45,866 --> 00:06:49,133
And for those of you who found
another team, make sure to compare.

177
00:06:49,133 --> 00:06:50,666
The steps that you did yourself

178
00:06:50,666 --> 00:06:53,800
with the steps that we did
in this tutorial to spot which.

179
00:06:53,800 --> 00:06:56,400
Difference led you to different results.

180
00:06:56,400 --> 00:06:58,866
You can always ask me
some questions in the Q&A.

181
00:06:58,866 --> 00:07:00,900
I'll be happy to help you with your model.

182
00:07:00,900 --> 00:07:04,566
But clearly the final team is either
the R&D spend or the R&D

183
00:07:04,566 --> 00:07:06,033
spend, plus the marketing spend.

184
00:07:06,033 --> 00:07:08,200
And that's where.
We also. Obtain in Python.

185
00:07:08,200 --> 00:07:09,600
And that's what you will also.

186
00:07:09,600 --> 00:07:13,533
Obtain in some other programing language
or machine learning package.

187
00:07:14,133 --> 00:07:15,900
So thank you. For watching this tutorial.

188
00:07:15,900 --> 00:07:17,266
Congratulations again.

189
00:07:17,266 --> 00:07:19,200
I hope you enjoyed doing this homework.

190
00:07:19,200 --> 00:07:21,866
You will have some other homeworks
in the other parts.

191
00:07:21,866 --> 00:07:26,266
So you'll definitely practice
which will help you shape your expertise.

192
00:07:26,266 --> 00:07:28,000
In machine learning.

193
00:07:28,000 --> 00:07:29,700
And speaking of machine learning,

194
00:07:29,700 --> 00:07:31,800
I look forward to seeing you
in the next tutorial.

195
00:07:31,800 --> 00:07:33,733
And until then, enjoy machine learning.