1
00:00:00,180 --> 00:00:06,270
Hello, all so in all the previous session, we have analyzed our data and we all have understand our

2
00:00:06,270 --> 00:00:07,830
data in a very proper way.

3
00:00:08,430 --> 00:00:14,190
This is exactly the trend from where my other guests are, what exactly the distribution of the prize

4
00:00:14,430 --> 00:00:20,090
and how exactly the prices of a room that I am in, which month my guest is higher.

5
00:00:20,250 --> 00:00:23,140
We all have analyzed this trend as well.

6
00:00:23,730 --> 00:00:27,550
So that's a time for your machine learning aspect.

7
00:00:27,870 --> 00:00:35,790
Now, the very first task with respect to your machine learning approach is exactly you have to select

8
00:00:36,150 --> 00:00:41,970
some important features using correlation concept for your machine learning model.

9
00:00:42,420 --> 00:00:49,110
So if you guys don't know what exactly the correlation, kindly follow my all the previous video that

10
00:00:49,110 --> 00:00:50,300
I have uploaded.

11
00:00:50,520 --> 00:00:58,150
So let me call ahead on my data so that I will get a quick overview of how exactly my data looks like.

12
00:00:58,500 --> 00:01:03,040
So this is exactly my data on which I have to find our correlations.

13
00:01:03,040 --> 00:01:08,560
So to find correlation, you have to just call a function which is exactly C Odawara.

14
00:01:08,880 --> 00:01:10,650
So just execute it.

15
00:01:10,770 --> 00:01:15,930
So it will give us this amazing metrics with respect to each and every feature.

16
00:01:15,940 --> 00:01:18,060
You have some correlation.

17
00:01:18,180 --> 00:01:24,740
And this is exactly like, of course, you will see with respect to this is cancer, to cancer.

18
00:01:24,840 --> 00:01:31,050
You have a coalition of one because it makes no sense at all having correlation between same features.

19
00:01:31,080 --> 00:01:36,400
Whereas with respect to is cancer, you have a coalition of zero point two nine.

20
00:01:36,630 --> 00:01:44,670
It means if this lead-time value increases, there is a twenty nine percent probability that this is

21
00:01:44,670 --> 00:01:46,200
cancer and also gets increased.

22
00:01:46,230 --> 00:01:49,710
That exactly is an exact meaning of this correlation.

23
00:01:50,160 --> 00:01:57,390
So now what I have to do, basically, I just need my correlation with respect to this is cancer.

24
00:01:57,600 --> 00:01:58,140
So far.

25
00:01:58,140 --> 00:02:00,660
This what we can do here?

26
00:02:00,660 --> 00:02:04,830
I'm just going to say data don't call correlation.

27
00:02:04,830 --> 00:02:07,860
And here I have to just paste over there.

28
00:02:07,920 --> 00:02:14,340
And if I'm going to execute it, it will hit on this amazing stage with respect to all the correlation

29
00:02:14,340 --> 00:02:21,540
value with respect to this is canceled because it's cancer is exactly my independent feature that we

30
00:02:21,540 --> 00:02:25,160
have to predict whether booking is going to be canceled or not.

31
00:02:25,470 --> 00:02:33,150
So we can easily figure out over here how all these variables are going to impact on this is canceled

32
00:02:33,150 --> 00:02:33,710
feature.

33
00:02:33,720 --> 00:02:40,320
It means how all these variables are going to impact how my bookings are going to cancel or not.

34
00:02:40,590 --> 00:02:42,830
So this is the exact meaning of this coalition.

35
00:02:43,230 --> 00:02:44,280
So let's see.

36
00:02:44,280 --> 00:02:46,730
This is exactly my very first over here.

37
00:02:46,770 --> 00:02:49,830
Let's say this is exactly my correlation.

38
00:02:49,830 --> 00:02:54,030
And after it now we have to simply pretend this coordination.

39
00:02:54,030 --> 00:02:54,540
That's it.

40
00:02:54,540 --> 00:03:00,330
And this is exactly the final correlation value that you have to take care of it.

41
00:03:00,570 --> 00:03:04,410
Let's say I have a shot this so far, this what we guys can do.

42
00:03:04,410 --> 00:03:07,740
Very first, let me just add to this call this correlation.

43
00:03:08,040 --> 00:03:16,080
And on this, if I'm going to call this absolute, because I have to neglect this negative and positive

44
00:03:16,080 --> 00:03:16,520
as well.

45
00:03:16,710 --> 00:03:20,940
So I have to just call this absolute and let me just execute all this.

46
00:03:21,210 --> 00:03:25,260
You will see all your positive and negative gas disappeared.

47
00:03:25,470 --> 00:03:32,790
And now on this, I have to just call this short on a score, Riluzole, just to execute all your stuff.

48
00:03:32,810 --> 00:03:36,480
So let's say I have to shorten in my descending order.

49
00:03:36,490 --> 00:03:38,730
So for this, I would say ascending.

50
00:03:38,970 --> 00:03:40,420
It goes to four.

51
00:03:40,470 --> 00:03:41,930
That set just execute.

52
00:03:41,940 --> 00:03:49,590
And it is your amazing steps over here that you really need to take care of it, from which you have

53
00:03:49,590 --> 00:03:53,520
to extract some meaningful and amazing insights.

54
00:03:53,580 --> 00:03:59,820
So from this correlation value, you can directly say that this time, this total number of special

55
00:03:59,820 --> 00:04:08,040
request, these required car, all these things are my most important features, that we have to take

56
00:04:08,040 --> 00:04:11,920
care of it because its coordination is definitely high.

57
00:04:12,090 --> 00:04:15,180
Let's say I have to understand this data is too.

58
00:04:15,360 --> 00:04:22,230
So for this, what we guys can do, let's say with respect to this dependent variable, I have to know

59
00:04:22,950 --> 00:04:31,560
with respect to this is concerned how many gas we have that have Check-Out that have canceled and that

60
00:04:31,710 --> 00:04:33,420
doesn't have any showboating.

61
00:04:33,870 --> 00:04:38,630
So for this what we can do, we can see DataDot will buy.

62
00:04:38,880 --> 00:04:46,070
So basically we have to group our data on the basis of this is underscore cancel.

63
00:04:46,080 --> 00:04:52,590
So I'm going to say I have to group on the basis of this one, then I have to accept this reservation

64
00:04:52,590 --> 00:04:59,570
is status, which is exactly a column, which is I would say reservation and a score is.

65
00:05:00,130 --> 00:05:07,740
And on this, I have to simply call my value, so I'm just going to say, well, you underscore council

66
00:05:07,870 --> 00:05:11,410
are just executed and it is this amazing.

67
00:05:11,730 --> 00:05:15,820
You will see when your booking isn't going to cancel.

68
00:05:16,300 --> 00:05:19,000
And reservationist, who does this check out?

69
00:05:19,000 --> 00:05:20,770
You have that much number of guest.

70
00:05:21,100 --> 00:05:27,720
And when your booking is going to cancel and your reservation status is canceled, then you have that

71
00:05:27,730 --> 00:05:33,040
much number of guests and it is no sure you have that much number of guests.

72
00:05:33,250 --> 00:05:35,660
So you can simply understand your data.

73
00:05:35,830 --> 00:05:41,110
Yeah, there are a higher number of customers there, a higher number of guests who are going to check

74
00:05:41,110 --> 00:05:41,550
out.

75
00:05:41,560 --> 00:05:50,050
Let's say I have to fetch my numerical features and my categorical features separately from all these

76
00:05:50,050 --> 00:05:50,650
features.

77
00:05:50,860 --> 00:05:54,160
So very first, I have to exclude some variables.

78
00:05:54,310 --> 00:05:57,970
So I'm going to do final list, let's say list on a score.

79
00:05:57,970 --> 00:05:58,840
Not so.

80
00:05:58,840 --> 00:06:07,090
The very first variable that I have to exclude in my modeling purposes is exactly my days in waiting

81
00:06:07,090 --> 00:06:07,450
list.

82
00:06:07,780 --> 00:06:15,190
So the question that will come across your mind why I have to exclude this, because this is that feature

83
00:06:15,460 --> 00:06:20,350
that is never going to contribute, whether a booking is going to cancel or not.

84
00:06:20,560 --> 00:06:22,560
And you can definitely see what here.

85
00:06:22,840 --> 00:06:24,720
It is also a very low correlation value.

86
00:06:24,770 --> 00:06:29,860
It means it is not going to impact that much on my model.

87
00:06:29,860 --> 00:06:32,410
So I have to basically excluded after it.

88
00:06:32,410 --> 00:06:37,730
We have to exclude, let's say, arrival, underscore date in a school year.

89
00:06:37,740 --> 00:06:39,940
So I have to exclude this one as well.

90
00:06:40,330 --> 00:06:44,350
After what I have to do now, I have to just execute it.

91
00:06:44,350 --> 00:06:50,890
Now we have to fetch how many numerical features we have or how many numerical columns we have.

92
00:06:51,310 --> 00:06:58,090
So for this, I'm just going to say for column in DataDot columns, I have to basically iterate on each

93
00:06:58,090 --> 00:06:58,810
and every column.

94
00:06:59,140 --> 00:07:01,180
Then I'm going to put a condition over here.

95
00:07:01,250 --> 00:07:06,820
Rather, the data type of that particular column must not be object.

96
00:07:06,820 --> 00:07:14,140
So I'm going to say data column dot data type, doesn't it, to object.

97
00:07:14,140 --> 00:07:15,720
That is exact meaning of this one.

98
00:07:16,060 --> 00:07:21,030
And if this commission will satisfy, I'm basically going to consider that column.

99
00:07:21,430 --> 00:07:24,370
But still, I have to let's say I have to exclude this one.

100
00:07:24,610 --> 00:07:33,670
So I'm going to add the condition as a. column, not in whatever list I have defined, which is exactly

101
00:07:33,670 --> 00:07:34,770
list and a score.

102
00:07:34,820 --> 00:07:40,210
Not so this is exactly my entire code of list comprehension.

103
00:07:40,450 --> 00:07:47,290
And if you guys are not that much comfortable with this list comprehension code, so you what you can

104
00:07:47,290 --> 00:07:50,680
do, you can say it is for data and columns.

105
00:07:50,680 --> 00:07:51,760
Let me show it there.

106
00:07:52,030 --> 00:07:53,290
So you have two very first.

107
00:07:53,290 --> 00:08:00,160
Add this condition after eight, you have to add this one and after it you have this condition and this

108
00:08:00,160 --> 00:08:00,370
one.

109
00:08:00,640 --> 00:08:04,030
So here I am going to say this is nothing but this one.

110
00:08:04,030 --> 00:08:09,690
And once I have all this stuff, then I am simply going to append in my list very first.

111
00:08:09,730 --> 00:08:11,110
I have to define some list.

112
00:08:11,320 --> 00:08:16,090
Let's say these are my let's say I'm going to define list as calls.

113
00:08:16,450 --> 00:08:22,390
So I'm going to say calls, dot, pen, whatever column I have.

114
00:08:22,750 --> 00:08:27,760
And now I have to simply print my course list.

115
00:08:28,060 --> 00:08:34,400
So it is basically for those one who are not that much confortable with this list comprehension code,

116
00:08:34,420 --> 00:08:37,930
because whenever you are going to form from real words, then add use.

117
00:08:38,200 --> 00:08:40,840
You have to always write your optimised code.

118
00:08:40,850 --> 00:08:45,070
You can't write down lines of code, one hundred lines of code.

119
00:08:45,190 --> 00:08:51,730
You have to always optimize your code because more the number of lines of code have more the resources

120
00:08:51,730 --> 00:08:52,560
it will take.

121
00:08:52,570 --> 00:08:59,910
So let's say I'm going to name this list as numerical underscore features just executed.

122
00:09:00,140 --> 00:09:02,990
Let's say I have to print my numerical features as well.

123
00:09:03,310 --> 00:09:10,180
So now you will see these are all the features that are exactly my numerical features.

124
00:09:10,270 --> 00:09:13,760
So on this feature, I can do a lot of people sitting.

125
00:09:13,780 --> 00:09:17,590
Let's say that there is some outliers in or not.

126
00:09:17,860 --> 00:09:21,320
So on these features, we can still do lots of people.

127
00:09:21,910 --> 00:09:29,110
And if you guys still not much comfortable with this code, you can definitely go ahead with this logic.

128
00:09:29,350 --> 00:09:36,190
But try to work on this list comprehension code, because once you will win in your industry use case,

129
00:09:36,340 --> 00:09:40,050
you have to always write such a optimized code.

130
00:09:40,060 --> 00:09:43,330
That's why I'm going to recommend you follow this approach.

131
00:09:43,690 --> 00:09:47,590
So let me cross the cell.

132
00:09:47,590 --> 00:09:54,520
And once I have all this stuff, what I have to do, let's say I just need my some categorical features

133
00:09:54,520 --> 00:09:54,910
as well.

134
00:09:54,910 --> 00:09:58,810
Let's say I have to exclude some categorical columns as well.

135
00:09:59,080 --> 00:09:59,470
So.

136
00:09:59,690 --> 00:10:05,380
Me, show me all the columns, so DataDot columns, these are exactly all the columns over here that

137
00:10:05,390 --> 00:10:08,710
make the final list very false and colonoscope not.

138
00:10:09,020 --> 00:10:16,880
And in this list, the very first is exactly my arrival on a score date and a score here, which is

139
00:10:16,880 --> 00:10:18,580
exactly this one.

140
00:10:18,590 --> 00:10:25,260
So I'm going to say this is the very first feature that I don't have to consider in my model building,

141
00:10:25,280 --> 00:10:29,930
because this is exactly that feature that now would impact your model that much.

142
00:10:30,180 --> 00:10:37,430
So the second feature is exactly your assigned room type, which is exactly this one.

143
00:10:37,430 --> 00:10:40,820
So I'm going to say this is exactly my second feature.

144
00:10:40,820 --> 00:10:47,210
After it, we have our booking changes column, which is exactly this one.

145
00:10:47,210 --> 00:10:54,080
So I'm going to say this is exactly my, let's say, third column that I have to exclude after what

146
00:10:54,080 --> 00:10:54,820
we have to do.

147
00:10:55,010 --> 00:11:00,620
We have some other columns, which is exactly my reiterations status.

148
00:11:00,620 --> 00:11:06,110
So I'm going to say this is nothing but my reservation is status after eight.

149
00:11:06,170 --> 00:11:12,950
What we have to do, we have something known as a country because country never impacts that much on

150
00:11:12,950 --> 00:11:14,540
my model building up.

151
00:11:14,540 --> 00:11:17,930
To some extent it will impact, but they are up to a greater extent.

152
00:11:17,930 --> 00:11:19,750
It will now impact that much.

153
00:11:20,180 --> 00:11:27,430
And the second column that I have to exclude is exactly my days in waiting list.

154
00:11:27,440 --> 00:11:31,160
So yeah, that will never impact that much.

155
00:11:31,310 --> 00:11:33,950
So I have to just place over there once.

156
00:11:33,950 --> 00:11:37,210
I will please let me just print my all these things.

157
00:11:37,520 --> 00:11:41,060
So these are all the columns that I have to exclude.

158
00:11:41,270 --> 00:11:45,080
So now what I have to do, I just need some categorical features.

159
00:11:45,080 --> 00:11:55,060
So far this I'm just going to say for column in data dot columns and I have to put a condition if data

160
00:11:55,220 --> 00:12:03,000
of column dot daytimes whosever column has a data type of object.

161
00:12:03,000 --> 00:12:09,950
So I'm going to say equally cause to object, if this condition will get satisfied, then only I am

162
00:12:09,950 --> 00:12:12,530
going to consider it in my list.

163
00:12:12,980 --> 00:12:17,930
And still you have to add one more condition and column.

164
00:12:18,290 --> 00:12:19,630
Not in it.

165
00:12:19,650 --> 00:12:24,800
Me, that column must not be in my cat on score, not list.

166
00:12:25,040 --> 00:12:29,060
So that is my entire Codal again list comprehension.

167
00:12:29,570 --> 00:12:36,890
So if you guys are not that much comfortable, you can definitely write that code that I have written

168
00:12:36,890 --> 00:12:42,790
over just one or two minutes before using my tools that I have shown you.

169
00:12:43,130 --> 00:12:50,480
So here I have to say, let's say here I'm going to say this is exactly my categorical Alesco features

170
00:12:50,480 --> 00:12:51,740
just executed.

171
00:12:52,070 --> 00:12:53,960
So now I have to simply print.

172
00:12:54,260 --> 00:12:59,720
So these are my all the categorical features that we have to take care of it.

173
00:12:59,720 --> 00:13:04,760
So hope you love the session and you find this session very valuable and helpful for you.

174
00:13:05,040 --> 00:13:05,720
Thank you.

175
00:13:05,750 --> 00:13:06,700
Have a nice day.

176
00:13:07,190 --> 00:13:08,060
Keep learning.

177
00:13:08,060 --> 00:13:08,990
Keep growing.

178
00:13:08,990 --> 00:13:09,830
Keep practicing.