1
00:00:01,200 --> 00:00:02,910
Now, let's build a few other models.

2
00:00:09,660 --> 00:00:18,840
That's to feature engineering and modify features of our dataset and build few other models, so deep

3
00:00:18,840 --> 00:00:24,610
learning neural network contains hardcoded data processing, feature extraction and feature engineering.

4
00:00:25,050 --> 00:00:30,240
Those who don't require much future engineering with deep learning models.

5
00:00:31,870 --> 00:00:38,340
So in order to get good accuracy with our regular machine learning models, we need to look at engineering,

6
00:00:39,250 --> 00:00:44,050
so let's modify our feature columns, so.

7
00:00:47,080 --> 00:00:49,030
Let's modify our age column first.

8
00:00:51,400 --> 00:00:55,600
So first, we're creating an empty list with the name.

9
00:00:56,740 --> 00:01:01,210
So here we are creating for loop and opening those values over age.

10
00:01:02,290 --> 00:01:06,290
So for I in range, land of date of.

11
00:01:08,870 --> 00:01:10,970
So we will run Lupa.

12
00:01:12,010 --> 00:01:13,530
To the left to hold it up, Adam.

13
00:01:14,810 --> 00:01:19,970
If your data from of age, I would use that particular rule.

14
00:01:22,220 --> 00:01:29,640
Less than or equal to 45 each dot append your legenda, youth to adult.

15
00:01:32,710 --> 00:01:40,860
So the age forty five, we're appealing that role in age as adult.

16
00:01:42,350 --> 00:01:51,080
And as is what senior, so if the aid is equal to or less than forty five, then it's ordered if not.

17
00:01:53,150 --> 00:01:56,180
Blitzer, Anderson and Anderson.

18
00:01:58,100 --> 00:01:59,980
Non-automotive I them once employed in.

19
00:02:04,150 --> 00:02:10,950
So here we will calculate months employed as months employed, plus years employed.

20
00:02:12,190 --> 00:02:17,520
So using this condition, we are saving that value in one Templer column.

21
00:02:19,890 --> 00:02:24,660
So for this, we are creating an entire list of the name employed.

22
00:02:25,920 --> 00:02:28,050
For I in range of data from.

23
00:02:28,980 --> 00:02:32,370
It's his goal to be a force once employed.

24
00:02:34,520 --> 00:02:44,450
I used that particular rule, plus the four years employed I, which is rule in goodwill, so suppose.

25
00:02:46,210 --> 00:02:47,560
We have this table.

26
00:02:49,140 --> 00:02:50,070
They've got this little.

27
00:02:52,080 --> 00:02:54,780
One, two, three, four, five, six.

28
00:02:55,290 --> 00:02:57,860
So first for I in range, land of the Earth.

29
00:02:57,870 --> 00:03:01,320
So this is the land of the earth for our children to be of.

30
00:03:02,490 --> 00:03:04,650
So here first our I will read this one.

31
00:03:06,400 --> 00:03:14,890
First rule, so it's difficult to deal for months in blind eye, which is months in blood cholesterol.

32
00:03:16,360 --> 00:03:17,350
It just this little.

33
00:03:21,410 --> 00:03:26,840
And we have all feature columns, so in this role, we have one simple column.

34
00:03:30,690 --> 00:03:32,880
Plus, days of years in blood column.

35
00:03:35,900 --> 00:03:36,120
Well.

36
00:03:37,210 --> 00:03:42,520
So these two columns, these two columns plus.

37
00:03:43,950 --> 00:03:44,480
Interruptible.

38
00:03:48,240 --> 00:03:49,560
Plus, an incredible.

39
00:03:50,520 --> 00:03:55,050
So once employed, Alice is under employed.

40
00:03:56,320 --> 00:04:05,460
And so we are doing this for all our roles and we are spending it in blood list.

41
00:04:10,290 --> 00:04:12,780
Little Anderson, a shift in attitude on the undersell.

42
00:04:14,500 --> 00:04:21,270
Here also in personal account, we are doing the same personal account month, less personal account

43
00:04:21,280 --> 00:04:26,080
here in Boothville, and we are bringing that to be list.

44
00:04:27,980 --> 00:04:28,840
Blitzer Anderson.

45
00:04:32,070 --> 00:04:33,240
Nonexplosive for the.

46
00:04:36,140 --> 00:04:44,380
So now let's calculate average of risk score so we have four risk, of course, risk to score three

47
00:04:45,200 --> 00:04:46,880
or four, original score five.

48
00:04:48,130 --> 00:04:50,980
So we are taking advantage of this for risk.

49
00:04:52,150 --> 00:04:57,670
So first, we are creating an empty list with the name average underscore risk, and yet we are creating

50
00:04:57,670 --> 00:05:01,300
a loop for I in range of div.

51
00:05:01,870 --> 00:05:07,180
It is going to be at its core to I, which is first column.

52
00:05:08,380 --> 00:05:16,800
The army arrived in the first column, so when we are hydrating with the first it all, so you're the

53
00:05:17,770 --> 00:05:23,980
first day of a rescue squad, three column, first from the rescue squad for column, first through.

54
00:05:25,330 --> 00:05:27,400
Deal of risk for Fuster.

55
00:05:28,550 --> 00:05:32,840
So we're adding all this and developing it before, so we're taking an average of.

56
00:05:33,850 --> 00:05:39,020
This Aristarchus course prefers to call them and appealing to average risk for.

57
00:05:41,260 --> 00:05:48,880
So when I used to do in this range, then we are doing the same for Rotu and we are bringing that.

58
00:05:50,360 --> 00:05:52,130
As average, it is for the.

59
00:05:53,080 --> 00:05:56,500
So similarly, we are doing it for quality.

60
00:05:57,620 --> 00:05:58,250
Here we have.

61
00:06:00,080 --> 00:06:02,030
Qualities, corridor and qualities good.

62
00:06:02,870 --> 00:06:03,590
So we are.

63
00:06:04,730 --> 00:06:13,190
Taking average of quality and civility and the quality list, let's run these totals.

64
00:06:18,580 --> 00:06:24,640
So now we have created so many feature columns, now let's convert all of those features, modified

65
00:06:24,640 --> 00:06:26,050
features to data from.

66
00:06:28,170 --> 00:06:33,960
So first will convert all the list into data from so ages, pages of data from each.

67
00:06:34,350 --> 00:06:35,820
So we have created a list.

68
00:06:36,120 --> 00:06:38,250
So we are converting that age list into data.

69
00:06:38,250 --> 00:06:42,450
From then we'll complete the converting employed list to employer data from.

70
00:06:44,110 --> 00:06:54,570
And then be a little to be for the average risk to average risk as debt of Adam and Eve to quality data

71
00:06:54,580 --> 00:06:54,880
for from.

72
00:06:56,100 --> 00:07:04,540
And then finally, we're creating the name feature and concatenating all this little of the transition.

73
00:07:05,160 --> 00:07:11,780
So if you have all these features but we don't have names for the columns and here we are trading column

74
00:07:11,880 --> 00:07:17,620
names, feature columns is equal to age and blood being risk and quality.

75
00:07:18,390 --> 00:07:22,500
So these are the column names for our new date of transition.

76
00:07:23,750 --> 00:07:26,570
Let's visualize our new feature data from.

77
00:07:28,570 --> 00:07:36,760
So here we have each peerce quality through each column is category, rather than seeing it for all

78
00:07:36,760 --> 00:07:37,990
other columns are in numbers.

79
00:07:38,000 --> 00:07:41,110
So let's convert over each column, the numbers.

80
00:07:42,310 --> 00:07:49,150
So we're creating a dummy variable and we're taking each column and we are studying it and we want to

81
00:07:49,150 --> 00:07:51,490
visualize this, that we do different.

82
00:07:53,180 --> 00:07:55,180
We will not head until.

83
00:07:56,890 --> 00:07:58,540
So here we have adult.

84
00:07:59,980 --> 00:08:04,400
So the only one we have stored age in numbers.

85
00:08:05,110 --> 00:08:12,760
So now let's remove our age column, which is in categorical format, which isn't saying you're an adult.

86
00:08:13,090 --> 00:08:18,120
Let's remove this age from our future date of rape and add this column.

87
00:08:19,780 --> 00:08:25,980
So first feature the drop age, they're dropping like the column and then feature is going to.

88
00:08:27,240 --> 00:08:34,610
We did the concatenate feature column feature data for them, and then we did offer them endless.

89
00:08:35,860 --> 00:08:41,650
Join this together and visualize the data from the writing feature that lets Handsell.

90
00:08:45,190 --> 00:08:46,540
So here we have convertor.

91
00:08:48,630 --> 00:08:51,260
All of it categorical features into numbers.

92
00:08:52,590 --> 00:08:53,740
Now, let's proceed further.

93
00:08:54,540 --> 00:08:55,680
Let's proceed further.

94
00:08:56,130 --> 00:09:01,770
Now let's update our features and labels after doing feature engineering.

95
00:09:04,330 --> 00:09:08,530
Let's create a variable with the dependent variable.

96
00:09:10,030 --> 00:09:12,250
Registering as a sign.

97
00:09:12,490 --> 00:09:16,030
So this is a string and we are saving this string in the.

98
00:09:18,260 --> 00:09:25,410
And here are creating a well with the name and the eye is called to the columns to.

99
00:09:26,710 --> 00:09:29,920
So, yes, we are storing all the column names.

100
00:09:30,970 --> 00:09:32,580
Of our data from.

101
00:09:35,720 --> 00:09:36,350
Run, Anderson.

102
00:09:38,190 --> 00:09:40,050
I will show you the value of andI.

103
00:09:43,050 --> 00:09:44,750
It's creating little.

104
00:09:51,610 --> 00:09:53,880
So it has all these different names.

105
00:09:55,660 --> 00:10:00,420
So now from this list, we need to remove this sign.

106
00:10:01,330 --> 00:10:08,400
So for that, you need to remove the B so deep we have to sign the string.

107
00:10:09,100 --> 00:10:11,360
So which will remove the underscore to sign the string.

108
00:10:11,650 --> 00:10:13,920
And we also need to remove a.

109
00:10:15,370 --> 00:10:19,420
So we are removing these two columnists from this list.

110
00:10:23,110 --> 00:10:30,700
Now, let's separate our X and Y, so X is according to data from us, and so we are selecting all these

111
00:10:30,700 --> 00:10:31,480
column names.

112
00:10:32,860 --> 00:10:34,090
In our feature column.

113
00:10:35,350 --> 00:10:44,060
In our features did of them and vice custody of BP, which is overdesign, so envie, which is our labels

114
00:10:44,320 --> 00:10:50,210
from we are selecting only one column and an X, we have all columns except our label column.

115
00:10:50,650 --> 00:10:51,730
It's Grandison.

116
00:10:54,220 --> 00:11:03,070
Now, let's concatenate our modified feature columns with our X, so feature these our newly created

117
00:11:03,080 --> 00:11:05,410
data for them with all modified features.

118
00:11:06,590 --> 00:11:09,190
And we realize this, that having extorted.

119
00:11:11,700 --> 00:11:12,470
Israelson.

120
00:11:15,180 --> 00:11:21,030
So here these are the default columns, each basically dual homeowner current address personal.

121
00:11:22,310 --> 00:11:23,570
It is five.

122
00:11:25,370 --> 00:11:25,790
So.

123
00:11:28,530 --> 00:11:35,760
Until this inquiry is last month, these are default columns from your employer to be the quality of

124
00:11:35,970 --> 00:11:37,410
the columns which we have created.

125
00:11:38,630 --> 00:11:45,290
So as we have already modified few columns and created new columns to remove unnecessary columns.

126
00:11:47,230 --> 00:11:56,800
So Exercycle do drop Labor's call to age, once employed, it employed personal account, personal five,

127
00:11:57,550 --> 00:12:01,330
two or three or go to five.

128
00:12:03,100 --> 00:12:12,180
If the court is called to disqualify soldiers to remove all this columnists, let's understand that

129
00:12:12,820 --> 00:12:14,300
let's we knew it.

130
00:12:19,770 --> 00:12:22,500
So here we are, basically all column and.

131
00:12:23,920 --> 00:12:29,460
All the columns are in numbers, but only this particular column is in category, which is in strength.

132
00:12:29,470 --> 00:12:32,270
So we need to convert this column numbers.

133
00:12:33,320 --> 00:12:38,990
So let's create a dummy variable, dummy variable with the name dummy to.

134
00:12:40,320 --> 00:12:41,520
For Piscatella column.

135
00:12:43,600 --> 00:12:51,580
So we did not get the Army's of Piscatella column and it in the middle and in the middle, let's drop

136
00:12:51,580 --> 00:12:52,330
by quickly.

137
00:12:54,440 --> 00:12:57,390
Because you realize this will not hit.

138
00:12:59,570 --> 00:13:03,330
So after dropping by weekly, we have monthly, semi, monthly and weekly.

139
00:13:04,040 --> 00:13:12,160
So now let's remove your name from our exit and add this to.

140
00:13:14,680 --> 00:13:22,300
So first, we are dropping Piscatella column from EXER and we are concatenating our EXER with our newly

141
00:13:22,300 --> 00:13:23,620
created data from the middle.

142
00:13:24,580 --> 00:13:26,780
Which has vandalism and weekly.

143
00:13:28,530 --> 00:13:32,290
And the this will not help, Anderson.

144
00:13:34,160 --> 00:13:36,080
So this is our final data from.

145
00:13:39,850 --> 00:13:41,470
Which is all numerical values.

146
00:13:43,150 --> 00:13:46,150
Now, let's just kill all our values with standards killer.

147
00:13:46,630 --> 00:13:51,160
So first we are importing from Eskil under preprocessing import.