﻿1
00:00:00,133 --> 00:00:00,966
Hello, my friends.

2
00:00:00,966 --> 00:00:02,400
All right, let's do this.

3
00:00:02,400 --> 00:00:04,466
Let's start this. Implementation.

4
00:00:04,466 --> 00:00:06,233
So the first step is, of.

5
00:00:06,233 --> 00:00:08,733
Course, to implement
the data. Preprocessing phase.

6
00:00:08,733 --> 00:00:10,200
And it will be slightly.

7
00:00:10,200 --> 00:00:11,300
Different than what we.

8
00:00:11,300 --> 00:00:14,733
Did before
because we will only use actually from.

9
00:00:14,733 --> 00:00:16,600
Our data preprocessing template.

10
00:00:16,600 --> 00:00:17,700
This code cell.

11
00:00:17,700 --> 00:00:19,266
Where we import, you know, the.

12
00:00:19,266 --> 00:00:20,466
Widely used libraries.

13
00:00:20,466 --> 00:00:22,466
Like numpy, matplotlib and vendors.

14
00:00:22,466 --> 00:00:23,266
And then only.

15
00:00:23,266 --> 00:00:25,266
This. Line of code, you know, to.

16
00:00:25,266 --> 00:00:27,133
Load the data set. And that's it.

17
00:00:27,133 --> 00:00:27,700
Then we won't.

18
00:00:27,700 --> 00:00:28,100
Create a.

19
00:00:28,100 --> 00:00:31,566
Matrix of features or dependent variable
vector because here we're doing a.

20
00:00:31,566 --> 00:00:33,900
Totally different thing
the association rule learning.

21
00:00:33,900 --> 00:00:35,166
And we don't have to. Split the.

22
00:00:35,166 --> 00:00:36,900
Data set into a training set and test.

23
00:00:36,900 --> 00:00:40,800
It, because we will directly learn
all the rules through the whole data set.

24
00:00:40,900 --> 00:00:41,633
Okay.

25
00:00:41,633 --> 00:00:43,100
So right now we just want.

26
00:00:43,100 --> 00:00:44,300
This first, you know.

27
00:00:44,300 --> 00:00:47,100
To import the main libraries.

28
00:00:47,100 --> 00:00:48,833
And then we'll. Get that.

29
00:00:48,833 --> 00:00:50,333
Line of code to import the.

30
00:00:50,333 --> 00:00:51,000
Data set.

31
00:00:51,000 --> 00:00:53,233
So here I'm creating a new code cell.

32
00:00:53,233 --> 00:00:54,666
I'm importing this.

33
00:00:54,666 --> 00:00:56,500
And then we can load the data set.

34
00:00:56,500 --> 00:00:59,166
However there is something important
you should know about.

35
00:00:59,166 --> 00:01:01,466
This a priori. Implementation.

36
00:01:01,466 --> 00:01:04,066
It is the fact that for the first. Time
we won't.

37
00:01:04,066 --> 00:01:05,466
Use scikit learn.

38
00:01:05,466 --> 00:01:08,433
Unfortunately,
the cyclone library doesn't.

39
00:01:08,433 --> 00:01:11,233
Include some classes or functions on a.

40
00:01:11,233 --> 00:01:11,866
Priory.

41
00:01:11,866 --> 00:01:14,633
Basically,
it doesn't include the a priori model.

42
00:01:14,633 --> 00:01:16,166
Therefore, we're not going to.

43
00:01:16,166 --> 00:01:17,433
Use scikit learn to.

44
00:01:17,433 --> 00:01:18,633
Train the model.

45
00:01:18,633 --> 00:01:20,233
We will actually use another.

46
00:01:20,233 --> 00:01:22,866
Library
which is called a priory. You know.

47
00:01:22,866 --> 00:01:25,000
priory.py is actually a.

48
00:01:25,000 --> 00:01:27,500
Python implementation containing all. The.

49
00:01:27,500 --> 00:01:29,633
Algorithm of the Priory model.

50
00:01:29,633 --> 00:01:30,166
And that's.

51
00:01:30,166 --> 00:01:30,633
What we.

52
00:01:30,633 --> 00:01:33,400
Will get, you know, and use to train our.

53
00:01:33,400 --> 00:01:34,633
Priory. Model here on the.

54
00:01:34,633 --> 00:01:37,766
Whole data set,
but exceptionally, because.

55
00:01:37,766 --> 00:01:38,633
You know. Google.

56
00:01:38,633 --> 00:01:40,266
Colab contains most of.

57
00:01:40,266 --> 00:01:42,966
The libraries and packages pre-installed.

58
00:01:42,966 --> 00:01:45,266
But here, exceptionally, Google Colab.

59
00:01:45,266 --> 00:01:46,500
Doesn't include. That.

60
00:01:46,500 --> 00:01:47,633
Appearing module.

61
00:01:47,633 --> 00:01:50,266
So we will actually have to install it.

62
00:01:50,266 --> 00:01:52,633
And this. Is very good that you see this
because you.

63
00:01:52,633 --> 00:01:56,866
Will encounter this situation
sometimes, you know, rarely but sometimes.

64
00:01:57,133 --> 00:01:59,400
And so you need to know how to install.

65
00:01:59,400 --> 00:02:01,333
A certain package or a certain library.

66
00:02:01,333 --> 00:02:03,133
From the web, you know, from online.

67
00:02:03,133 --> 00:02:06,300
Because what we'll do now is
we will enter a command, you know, Pip

68
00:02:06,300 --> 00:02:07,266
command, which.

69
00:02:07,266 --> 00:02:09,533
Will download first the binary file.

70
00:02:09,533 --> 00:02:11,466
From the internet, you know, through link.

71
00:02:11,466 --> 00:02:14,200
And then we'll install it inside. This.

72
00:02:14,200 --> 00:02:15,433
Particular notebook.

73
00:02:15,433 --> 00:02:15,866
All right.

74
00:02:15,866 --> 00:02:17,566
Let me to. Show you let's do this.

75
00:02:17,566 --> 00:02:19,766
So we're. Actually
going to implement this.

76
00:02:19,766 --> 00:02:22,000
Right before you. Know
importing the main. Libraries.

77
00:02:22,000 --> 00:02:25,166
Usually the packages are installed
as a first step.

78
00:02:25,533 --> 00:02:28,666
And to do this
we need to start with an exclamation mark.

79
00:02:28,800 --> 00:02:31,100
Then Pip
you know that's the pip command that.

80
00:02:31,100 --> 00:02:33,366
Allows us to. Install. Packages.

81
00:02:33,366 --> 00:02:34,600
Then very simply.

82
00:02:34,600 --> 00:02:35,100
Well speaking.

83
00:02:35,100 --> 00:02:37,466
Of installing packages,
the next thing to enter here.

84
00:02:37,466 --> 00:02:38,800
Is install.

85
00:02:38,800 --> 00:02:40,933
And then you simply enter the name of.

86
00:02:40,933 --> 00:02:42,966
That library or module.

87
00:02:42,966 --> 00:02:44,266
That you want to install.

88
00:02:44,266 --> 00:02:47,100
And in our case. As we said it is a fire.

89
00:02:47,100 --> 00:02:47,700
Be careful.

90
00:02:47,700 --> 00:02:48,200
It is not a.

91
00:02:48,200 --> 00:02:53,266
Priory,
it is a fire like that AP or AI okay.

92
00:02:53,566 --> 00:02:56,933
So exclamation mark pip, install a binary.

93
00:02:57,366 --> 00:02:59,166
All right.
And now let me show. You what it does.

94
00:02:59,166 --> 00:03:01,733
As I said it's.
Going to download it. First.

95
00:03:01,733 --> 00:03:04,166
And then it will install it inside a.

96
00:03:04,166 --> 00:03:06,300
Notebook. So here it is.

97
00:03:06,300 --> 00:03:08,100
Finding it on the web.

98
00:03:08,100 --> 00:03:09,633
And very soon we will see in the.

99
00:03:09,633 --> 00:03:11,666
Output the download.

100
00:03:11,666 --> 00:03:13,833
All right there
we go. The download is starting now.

101
00:03:14,800 --> 00:03:15,733
collecting it by.

102
00:03:15,733 --> 00:03:16,733
Downloading.

103
00:03:16,733 --> 00:03:20,833
It. From the link
and then installing it and then, you know.

104
00:03:20,833 --> 00:03:22,400
Successfully. Installed. A binary.

105
00:03:22,400 --> 00:03:24,533
See it just takes a few seconds.

106
00:03:24,533 --> 00:03:25,766
And this is the version of this.

107
00:03:25,766 --> 00:03:28,166
Binary package. 1.1.2.

108
00:03:28,166 --> 00:03:30,166
Maybe you'll get a different one,
you know, if you.

109
00:03:30,166 --> 00:03:33,166
Take this course way
after I recorded this tutorial.

110
00:03:33,266 --> 00:03:33,933
All right. So good.

111
00:03:33,933 --> 00:03:35,566
Now you know how to install. A package.

112
00:03:35,566 --> 00:03:36,966
Within Google. Colab.

113
00:03:36,966 --> 00:03:38,700
But don't worry, most of the.

114
00:03:38,700 --> 00:03:41,700
Packages and including the deep learning
one like TensorFlow.

115
00:03:41,800 --> 00:03:44,033
Are already pre-installed.

116
00:03:44,033 --> 00:03:45,500
Okay, good.

117
00:03:45,500 --> 00:03:48,400
So now well let's import
the libraries now.

118
00:03:48,400 --> 00:03:49,633
As we're here.

119
00:03:49,633 --> 00:03:52,066
And now let's implement the.
Data preprocessing phase.

120
00:03:52,066 --> 00:03:53,866
So let's create a new code cell.

121
00:03:53,866 --> 00:03:54,433
Then as.

122
00:03:54,433 --> 00:03:57,433
We said we will just get that first.

123
00:03:57,433 --> 00:04:00,000
Line of importing the data
sets in our data.

124
00:04:00,000 --> 00:04:02,400
Preprocessing. Template.

125
00:04:02,400 --> 00:04:04,800
And we're going to paste. That.

126
00:04:04,800 --> 00:04:06,833
Right here in our copy.

127
00:04:06,833 --> 00:04:07,500
And that's it.

128
00:04:07,500 --> 00:04:09,133
That's all we'll get from the template.

129
00:04:09,133 --> 00:04:12,566
And now we will adapt
and make it 100% customized.

130
00:04:12,566 --> 00:04:14,733
To the primary model.

131
00:04:14,733 --> 00:04:15,033
All right.

132
00:04:15,033 --> 00:04:16,866
So first let's just replace.

133
00:04:16,866 --> 00:04:18,233
The name. Of the data set here by.

134
00:04:18,233 --> 00:04:21,233
The right name which is remember. Market.

135
00:04:21,833 --> 00:04:22,600
Underscore.

136
00:04:22,600 --> 00:04:26,200
Basket underscore. Optimization

137
00:04:27,200 --> 00:04:27,866
okay.

138
00:04:27,866 --> 00:04:29,466
And now that. We did this let's.

139
00:04:29,466 --> 00:04:31,300
Think of the right thing to do.

140
00:04:31,300 --> 00:04:34,000
Which is to upload the data set here.

141
00:04:34,000 --> 00:04:34,666
Oh good.

142
00:04:34,666 --> 00:04:36,633
So let's click upload.

143
00:04:36,633 --> 00:04:38,700
All right then make sure to fine.

144
00:04:38,700 --> 00:04:41,366
You know that machine. Learning
is that codes. And data sets.

145
00:04:41,366 --> 00:04:44,166
Folder on your. Machine.
And if you haven't downloaded. Yet.

146
00:04:44,166 --> 00:04:44,633
You can.

147
00:04:44,633 --> 00:04:47,700
Download it right before this tutorial
at the bottom of the article.

148
00:04:47,933 --> 00:04:48,300
All right.

149
00:04:48,300 --> 00:04:51,866
So let's go inside
and let's go to part five Association.

150
00:04:51,866 --> 00:04:52,733
Rule learning.

151
00:04:52,733 --> 00:04:55,533
Then Priory then. Python.
And there we. Go.

152
00:04:55,533 --> 00:04:58,700
Let's select market basket optimization.

153
00:04:59,100 --> 00:05:01,466
Let's click open. Click okay.

154
00:05:01,466 --> 00:05:03,266
And it. Will load it.

155
00:05:03,266 --> 00:05:04,966
And there we go. We have it.

156
00:05:04,966 --> 00:05:07,300
Don't try to open it
because it is actually very big.

157
00:05:07,300 --> 00:05:08,966
And Google Colab won't allow you.

158
00:05:08,966 --> 00:05:12,333
But you know if you want to have a
look at it, you can go back to the folder.

159
00:05:12,566 --> 00:05:13,600
And, you know.

160
00:05:13,600 --> 00:05:15,400
Just double click it here, right.

161
00:05:15,400 --> 00:05:18,500
And you can see it very well
and see all the transaction okay.

162
00:05:18,500 --> 00:05:21,166
So we'll actually leave it like that
in case we want to see it.

163
00:05:21,166 --> 00:05:22,200
And now. Back to.

164
00:05:22,200 --> 00:05:23,466
Our implementation.

165
00:05:23,466 --> 00:05:25,233
Let's see what we. Have to do. Next.

166
00:05:25,233 --> 00:05:28,300
So we already loaded the data set as.

167
00:05:28,300 --> 00:05:29,800
A pandas dataframe.

168
00:05:29,800 --> 00:05:32,466
However
we have to be careful with something.

169
00:05:32,466 --> 00:05:33,366
Look at the data set.

170
00:05:33,366 --> 00:05:35,366
Again and notice something.

171
00:05:35,366 --> 00:05:38,000
Notice that you know on our.
Previous data set.

172
00:05:38,000 --> 00:05:39,333
The first row here.

173
00:05:39,333 --> 00:05:42,000
Of this index one here contained actually.

174
00:05:42,000 --> 00:05:44,733
Before. The names of the. Columns, right.

175
00:05:44,733 --> 00:05:47,733
For example,
if we take our social network ads data.

176
00:05:47,733 --> 00:05:48,900
Set in. Part three.

177
00:05:48,900 --> 00:05:50,866
Classification, we'll remember this.

178
00:05:50,866 --> 00:05:53,366
First row here had first the age.

179
00:05:53,366 --> 00:05:54,700
The name of the first column.

180
00:05:54,700 --> 00:05:58,366
And then the estimated salary,
which was the name of the second column,

181
00:05:58,633 --> 00:06:01,466
and then the name of the dependent
variable purchased.

182
00:06:01,466 --> 00:06:04,133
And here in this. Particular data set.

183
00:06:04,133 --> 00:06:06,833
Well, you know,
we don't have names of columns because.

184
00:06:06,833 --> 00:06:08,433
Each of these elements.

185
00:06:08,433 --> 00:06:10,000
Here correspond to just.

186
00:06:10,000 --> 00:06:11,066
Different products. Okay.

187
00:06:11,066 --> 00:06:13,800
So there was no need to name
any of these columns.

188
00:06:13,800 --> 00:06:16,700
And therefore here
we need to add something in our.

189
00:06:16,700 --> 00:06:19,000
Read underscore CSV. Function to.

190
00:06:19,000 --> 00:06:21,033
Tell that the first row.

191
00:06:21,033 --> 00:06:24,233
Doesn't contain the names of the column,
because if we don't say this.

192
00:06:24,233 --> 00:06:25,500
Well, actually it. Will not.

193
00:06:25,500 --> 00:06:27,500
Take that first. Transaction.

194
00:06:27,500 --> 00:06:28,533
Because it will. Think.

195
00:06:28,533 --> 00:06:31,100
That this first row. Are just names of.
The columns.

196
00:06:31,100 --> 00:06:32,500
We have to specify. This and the.

197
00:06:32,500 --> 00:06:34,633
Way to specify this. And it's good that.

198
00:06:34,633 --> 00:06:36,566
You know, this trick in pandas.

199
00:06:36,566 --> 00:06:37,033
Is just.

200
00:06:37,033 --> 00:06:39,766
Add another parameter
in this read underscore.

201
00:06:39,766 --> 00:06:40,800
Csb. Function.

202
00:06:40,800 --> 00:06:43,833
Which is header, and we set it equal.

203
00:06:43,833 --> 00:06:45,500
To num. Right.

204
00:06:45,500 --> 00:06:49,466
So that we can specify that indeed
there are no headers.

205
00:06:49,466 --> 00:06:51,766
Meaning no column names. Okay.

206
00:06:51,766 --> 00:06:52,800
That's what it means.

207
00:06:52,800 --> 00:06:54,600
So now you know this.
And therefore. It will.

208
00:06:54,600 --> 00:06:56,366
Take into account, you know, it will.

209
00:06:56,366 --> 00:06:59,066
Take the first row of this market basket.

210
00:06:59,066 --> 00:07:01,966
Optimization data set
containing all the transactions.

211
00:07:01,966 --> 00:07:04,500
And we'll do this.
It will take the first transaction.

212
00:07:04,500 --> 00:07:07,466
All right. Good.
So that was the first step.

213
00:07:07,466 --> 00:07:10,366
But then we actually. Have another.
Big step.

214
00:07:10,366 --> 00:07:10,766
And that.

215
00:07:10,766 --> 00:07:14,066
Has to do with the fact
that when we will train the a.

216
00:07:14,066 --> 00:07:15,666
Priori model on the data.

217
00:07:15,666 --> 00:07:17,200
Set, well we will use a.

218
00:07:17,200 --> 00:07:20,100
Certain function
that. Is actually called a priory.

219
00:07:20,100 --> 00:07:21,100
And of course that function.

220
00:07:21,100 --> 00:07:24,100
Will take. As input. The data set.

221
00:07:24,300 --> 00:07:25,200
But the thing is.

222
00:07:25,200 --> 00:07:28,200
That it expects this data set to have a.

223
00:07:28,200 --> 00:07:33,233
Certain format, and that format is
unfortunately not append this data frame.

224
00:07:33,433 --> 00:07:37,933
And therefore we have to recreate the data
set, you know, from this original.

225
00:07:37,933 --> 00:07:40,066
Pandas dataframe so that it can have.

226
00:07:40,066 --> 00:07:43,000
This format expected by the a priori.

227
00:07:43,000 --> 00:07:43,666
Function.

228
00:07:43,666 --> 00:07:46,100
Which will. Train
the Priory model on the whole.

229
00:07:46,100 --> 00:07:46,900
Data set.

230
00:07:46,900 --> 00:07:49,133
And so now the question is
what is this format?

231
00:07:49,133 --> 00:07:52,400
Well, that format is simply
a list of transactions.

232
00:07:52,400 --> 00:07:54,266
You know, instead of having the data set.

233
00:07:54,266 --> 00:07:56,233
As a. Pandas dataframe, we want to.

234
00:07:56,233 --> 00:07:58,733
Have the data
set as a list. Of transactions.

235
00:07:58,733 --> 00:08:00,733
You know, the transactions. Listed.

236
00:08:00,733 --> 00:08:01,666
One by one.

237
00:08:01,666 --> 00:08:04,666
With the different products
purchased by the customers.

238
00:08:05,100 --> 00:08:06,833
And that's exactly what we have to do now.

239
00:08:06,833 --> 00:08:08,133
We have to. Create that list.

240
00:08:08,133 --> 00:08:10,033
And the first.
Step to create that list is.

241
00:08:10,033 --> 00:08:11,100
Actually to initialize.

242
00:08:11,100 --> 00:08:13,600
It as. An empty. List.
Because then in order.

243
00:08:13,600 --> 00:08:16,266
To create that list,
we will have to populate it with.

244
00:08:16,266 --> 00:08:18,966
The different transactions of.
The data set.

245
00:08:18,966 --> 00:08:19,633
And the way we.

246
00:08:19,633 --> 00:08:21,900
Will do this is of course with a for loop.

247
00:08:21,900 --> 00:08:22,866
You know, for loop.

248
00:08:22,866 --> 00:08:26,133
Iterating all the 7500 transactions of.

249
00:08:26,133 --> 00:08:29,000
The data set so as to populate it. Okay.

250
00:08:29,000 --> 00:08:29,700
So let's do this.

251
00:08:29,700 --> 00:08:32,466
First let's. Initialize. That list.

252
00:08:32,466 --> 00:08:35,033
We will call it. Transactions and we will.

253
00:08:35,033 --> 00:08:37,400
Initialize it as. An empty list.

254
00:08:37,400 --> 00:08:37,733
All right.

255
00:08:37,733 --> 00:08:40,300
So so far it. Is just an empty list.
And now.

256
00:08:40,300 --> 00:08:44,133
We will start for loop
to populate this list of.

257
00:08:44,133 --> 00:08:45,800
Transactions with. All.

258
00:08:45,800 --> 00:08:48,866
The transactions contained in that pandas
DataFrame.

259
00:08:48,966 --> 00:08:50,166
Data set.

260
00:08:50,166 --> 00:08:50,633
Okay.

261
00:08:50,633 --> 00:08:52,733
So as we make that loop you.

262
00:08:52,733 --> 00:08:54,033
Will totally understand.

263
00:08:54,033 --> 00:08:55,233
What we we're doing.

264
00:08:55,233 --> 00:08:56,266
So for.

265
00:08:56,266 --> 00:08:58,900
Then. Classic iterated variable I.

266
00:08:58,900 --> 00:08:59,400
You know, which.

267
00:08:59,400 --> 00:09:02,566
Will take all the values from zero to.

268
00:09:02,566 --> 00:09:04,266
7500.

269
00:09:04,266 --> 00:09:07,266
But remember that
the upper bound of a range is excluded.

270
00:09:07,300 --> 00:09:10,533
So we will actually have to go
up. To 7501.

271
00:09:10,833 --> 00:09:13,833
Therefore here
this iterated variable will go.

272
00:09:13,933 --> 00:09:16,766
In the range from.

273
00:09:16,766 --> 00:09:18,366
Zero to.

274
00:09:18,366 --> 00:09:21,566
7501. All right.

275
00:09:21,600 --> 00:09:25,700
They're actually 7501 transactions

276
00:09:25,700 --> 00:09:28,766
because we start from zero and not 7500.

277
00:09:28,766 --> 00:09:31,200
Right. We can check it out very quickly.

278
00:09:31,200 --> 00:09:32,733
You know here we start from one.

279
00:09:32,733 --> 00:09:36,000
And when we scroll down
we go down to let's see.

280
00:09:36,000 --> 00:09:39,166
Yes 7501. So that's the exact.

281
00:09:39,333 --> 00:09:40,700
Number of transactions.

282
00:09:40,700 --> 00:09:41,400
All right.

283
00:09:41,400 --> 00:09:42,533
So all good.

284
00:09:42,533 --> 00:09:43,333
For the range.

285
00:09:43,333 --> 00:09:46,333
Now don't forget the column.

286
00:09:46,500 --> 00:09:48,700
And now we begin the for loop.

287
00:09:48,700 --> 00:09:51,800
And well you know building
this list of transactions is very easy.

288
00:09:51,800 --> 00:09:53,700
We will just use the append.

289
00:09:53,700 --> 00:09:56,400
Function which means add
and which will simply.

290
00:09:56,400 --> 00:09:58,366
You know,
add the. Different transactions of.

291
00:09:58,366 --> 00:09:59,966
The dataset. One by one.

292
00:09:59,966 --> 00:10:02,700
All right.
It's a very classic way to build a list.

293
00:10:02,700 --> 00:10:03,400
You know, you use.

294
00:10:03,400 --> 00:10:06,266
The append function
to add your different elements one by one.

295
00:10:06,266 --> 00:10:07,033
So let me.

296
00:10:07,033 --> 00:10:08,500
Show this to you very simply.

297
00:10:08,500 --> 00:10:08,866
What we need.

298
00:10:08,866 --> 00:10:10,200
To do here is take.

299
00:10:10,200 --> 00:10:13,133
Our transactions. List, okay.

300
00:10:13,133 --> 00:10:14,400
From which we're going to.

301
00:10:14,400 --> 00:10:15,033
Call this.

302
00:10:15,033 --> 00:10:17,933
Append function, which. Is a prebuilt.

303
00:10:17,933 --> 00:10:19,800
Function of a Python list.

304
00:10:19,800 --> 00:10:20,900
You know, Python has.

305
00:10:20,900 --> 00:10:21,733
All these pre-built.

306
00:10:21,733 --> 00:10:22,633
Functions.

307
00:10:22,633 --> 00:10:23,400
And inside.

308
00:10:23,400 --> 00:10:25,800
Well we're. Going to add the transaction.

309
00:10:25,800 --> 00:10:27,300
But we have to add. It in.

310
00:10:27,300 --> 00:10:28,800
A pair of square brackets.

311
00:10:28,800 --> 00:10:30,133
Because it's will contain, you know,

312
00:10:30,133 --> 00:10:32,100
all the different elements,
you know, all the different.

313
00:10:32,100 --> 00:10:33,900
Products purchased by.

314
00:10:33,900 --> 00:10:36,666
The underlying customer
in the transaction.

315
00:10:36,666 --> 00:10:37,800
And this transaction must.

316
00:10:37,800 --> 00:10:40,233
Be created as a list of. Products.

317
00:10:40,233 --> 00:10:40,800
And that's why.

318
00:10:40,800 --> 00:10:43,333
We have these new pair of square brackets.

319
00:10:43,333 --> 00:10:45,866
To make that transaction
a list. Of products.

320
00:10:45,866 --> 00:10:47,066
So in the end we're.

321
00:10:47,066 --> 00:10:49,166
Actually creating a list of lists.

322
00:10:49,166 --> 00:10:51,033
You know, each transaction in the big.

323
00:10:51,033 --> 00:10:54,000
List. Of transactions
is. Actually a list. All right.

324
00:10:54,000 --> 00:10:56,033
And so now there's a new little trick.

325
00:10:56,033 --> 00:10:58,233
But which is good. That, you know.

326
00:10:58,233 --> 00:11:00,900
It is a single row for loop.

327
00:11:00,900 --> 00:11:01,800
Because now actually.

328
00:11:01,800 --> 00:11:04,066
We need to do a second for loop to.

329
00:11:04,066 --> 00:11:05,066
You know.

330
00:11:05,066 --> 00:11:08,866
Get all the products of each transaction.

331
00:11:08,866 --> 00:11:10,200
So let me scroll back up.

332
00:11:10,200 --> 00:11:11,700
You know, the. First for loop will go.

333
00:11:11,700 --> 00:11:14,766
From this transaction
to this one to this one, to this one,

334
00:11:14,766 --> 00:11:18,300
down to the last one at the bottom 7501.

335
00:11:18,600 --> 00:11:19,200
But then we need to.

336
00:11:19,200 --> 00:11:21,533
Do a second for loop, which will loop.

337
00:11:21,533 --> 00:11:24,400
Over the different products
in each transaction.

338
00:11:24,400 --> 00:11:26,566
So you know it will. Add this product.

339
00:11:26,566 --> 00:11:28,533
Then this one,
then this one, then this one.

340
00:11:28,533 --> 00:11:31,833
And since the maximum number of products
in one basket.

341
00:11:31,833 --> 00:11:33,766
Is actually 20. You know.

342
00:11:33,766 --> 00:11:35,366
I put this first.

343
00:11:35,366 --> 00:11:37,166
Transaction at the. Top so that we.

344
00:11:37,166 --> 00:11:39,766
Can have the maximum. Size of the basket.

345
00:11:39,766 --> 00:11:40,800
Which is 20.

346
00:11:40,800 --> 00:11:43,100
And therefore what we'll do
now is the second for.

347
00:11:43,100 --> 00:11:46,566
Loop that will iterate from zero to. 20.

348
00:11:46,900 --> 00:11:48,900
It will start from zero here.

349
00:11:48,900 --> 00:11:50,833
You know.
That's the index. Of the first. Column.

350
00:11:50,833 --> 00:11:53,933
And then it will iterate to this one,
then to this one, then to this one

351
00:11:53,933 --> 00:11:57,266
up to the final index, which is index 19.

352
00:11:57,266 --> 00:12:00,333
Actually, because it starts from zero
there are 20 columns.

353
00:12:00,466 --> 00:12:02,833
So it goes. Up to the index 19.

354
00:12:02,833 --> 00:12:03,700
Okay. So that's the.

355
00:12:03,700 --> 00:12:04,566
Second for loop.

356
00:12:04,566 --> 00:12:05,400
And that's what.

357
00:12:05,400 --> 00:12:06,900
We'll do right away.

358
00:12:06,900 --> 00:12:07,866
Inside this.

359
00:12:07,866 --> 00:12:09,466
Append function to.

360
00:12:09,466 --> 00:12:10,366
Add all these.

361
00:12:10,366 --> 00:12:12,933
Different elements in each transaction.

362
00:12:12,933 --> 00:12:17,066
And for the transactions that don't have
20 elements, well that's totally fine.

363
00:12:17,300 --> 00:12:19,366
We will still iterate up to the.

364
00:12:19,366 --> 00:12:21,466
End, you know, up to the index 19.

365
00:12:21,466 --> 00:12:23,133
But we will just populate the list.

366
00:12:23,133 --> 00:12:26,833
With nones,
you know, Nans values which mean empty,

367
00:12:26,833 --> 00:12:30,900
so that our model will understand
that the transaction here only has.

368
00:12:30,900 --> 00:12:32,800
Three products. Okay.

369
00:12:32,800 --> 00:12:35,100
So we can totally iterate up to 20.

370
00:12:35,100 --> 00:12:37,566
And therefore let's start this second.

371
00:12:37,566 --> 00:12:38,500
For loop. Here.

372
00:12:38,500 --> 00:12:41,133
For now we need to take another.

373
00:12:41,133 --> 00:12:44,766
Iterated variable,
which we'll call naturally j for.

374
00:12:44,766 --> 00:12:47,500
J in the range from.

375
00:12:47,500 --> 00:12:49,366
Zero to be careful.

376
00:12:49,366 --> 00:12:51,700
Not 19, but 20.

377
00:12:51,700 --> 00:12:54,500
Because the. Upper bound
is once again. Excluded.

378
00:12:54,500 --> 00:12:57,233
So from 0 to 20 then what we will do.

379
00:12:57,233 --> 00:12:58,200
And that's.

380
00:12:58,200 --> 00:13:00,266
The syntax in a single.

381
00:13:00,266 --> 00:13:03,600
Row for loop we have to do what we want
to do with this.

382
00:13:03,600 --> 00:13:04,166
For loop

383
00:13:04,166 --> 00:13:08,966
at the beginning, you know, before the for
and what we want to do is just get that.

384
00:13:08,966 --> 00:13:10,733
Product in the transaction.

385
00:13:10,733 --> 00:13:12,733
Well from the index. Zero to.

386
00:13:12,733 --> 00:13:15,066
The index 19 and to. Access this product.

387
00:13:15,066 --> 00:13:17,500
Well we will of course use our data set.

388
00:13:17,500 --> 00:13:19,333
And play with the right indexes.

389
00:13:19,333 --> 00:13:20,733
To get the right product.

390
00:13:20,733 --> 00:13:21,033
All right.

391
00:13:21,033 --> 00:13:25,233
So here we need to call first
the data set okay.

392
00:13:25,600 --> 00:13:28,100
And now let's add some. Brackets.

393
00:13:28,100 --> 00:13:29,833
To add you know the index. Of the row.

394
00:13:29,833 --> 00:13:32,966
And the index of the column
that contains that product.

395
00:13:32,966 --> 00:13:35,333
We want to include in this transaction.

396
00:13:35,333 --> 00:13:36,566
So first according to you.

397
00:13:36,566 --> 00:13:38,566
What will be the index of the row.

398
00:13:38,566 --> 00:13:41,433
Well that will be of course I because I

399
00:13:41,433 --> 00:13:44,566
iterate
through all the rows of the data set.

400
00:13:44,800 --> 00:13:46,633
And therefore now we're dealing actually.

401
00:13:46,633 --> 00:13:47,700
With that particular.

402
00:13:47,700 --> 00:13:50,166
Transaction of index I. Which were.

403
00:13:50,166 --> 00:13:51,800
Appending into that transactions.

404
00:13:51,800 --> 00:13:55,766
List, and therefore the row of the data
set we are at.

405
00:13:55,766 --> 00:13:56,600
Right now.

406
00:13:56,600 --> 00:13:58,100
Is exactly I.

407
00:13:58,100 --> 00:14:00,466
Okay, I will start from zero.

408
00:14:00,466 --> 00:14:03,000
It will first get that transaction.

409
00:14:03,000 --> 00:14:04,500
Then it will be equal to one.

410
00:14:04,500 --> 00:14:05,166
So it will get.

411
00:14:05,166 --> 00:14:08,400
That transaction
and then this one and this one etc..

412
00:14:08,600 --> 00:14:09,166
All right.

413
00:14:09,166 --> 00:14:12,166
And so now we're dealing with a particular
transaction of a particular row.

414
00:14:12,300 --> 00:14:12,833
And that's.

415
00:14:12,833 --> 00:14:14,666
Exactly the index of the.

416
00:14:14,666 --> 00:14:16,233
Data set we need to input here.

417
00:14:16,233 --> 00:14:19,433
And now for the column
which index you think it's going to be.

418
00:14:19,700 --> 00:14:22,400
Well that will be of course
that. Index, you know, which.

419
00:14:22,400 --> 00:14:24,266
Iterates through all the.

420
00:14:24,266 --> 00:14:25,933
Columns of. Actually.

421
00:14:25,933 --> 00:14:27,833
The transaction. Right.

422
00:14:27,833 --> 00:14:29,500
J will go from. Zero.

423
00:14:29,500 --> 00:14:32,700
To 1 to 2. To 3 to 4 up. To 19.

424
00:14:32,833 --> 00:14:34,000
And therefore. Here.

425
00:14:34,000 --> 00:14:36,000
The index will one for a column is of.

426
00:14:36,000 --> 00:14:37,733
Course J.

427
00:14:37,733 --> 00:14:38,366
All right.

428
00:14:38,366 --> 00:14:39,666
So very good.

429
00:14:39,666 --> 00:14:42,000
But now we need to add one final thing.

430
00:14:42,000 --> 00:14:43,200
You know unfortunately we can.

431
00:14:43,200 --> 00:14:46,366
Not directly access the cell of row.

432
00:14:46,666 --> 00:14:49,133
And column
J in the. Data set. In order to.

433
00:14:49,133 --> 00:14:53,200
Access the cell
we just need to add here dot values.

434
00:14:53,466 --> 00:14:57,100
Because that's the part of that data
set structure.

435
00:14:57,100 --> 00:14:59,866
You know,
this is an advanced. Structure by. Pandas.

436
00:14:59,866 --> 00:15:02,300
That allow us to get access to the cells.
All right.

437
00:15:02,300 --> 00:15:04,000
So good. That, you know. This.

438
00:15:04,000 --> 00:15:05,900
And now we're almost ready.

439
00:15:05,900 --> 00:15:07,800
We almost. Have everything.

440
00:15:07,800 --> 00:15:09,200
The only thing that's left

441
00:15:09,200 --> 00:15:13,233
has again, to do with, you know,
a certain expectation of the future.

442
00:15:13,233 --> 00:15:16,233
A primary function which we'll use
to train our primary model.

443
00:15:16,500 --> 00:15:18,300
It is the fact that all the.

444
00:15:18,300 --> 00:15:19,666
Elements in the list.

445
00:15:19,666 --> 00:15:21,600
Must be strings.

446
00:15:21,600 --> 00:15:23,433
They must. Be strings. Otherwise.

447
00:15:23,433 --> 00:15:26,433
So a primary model
won't be able to learn the rules.

448
00:15:26,700 --> 00:15:27,300
And to make.

449
00:15:27,300 --> 00:15:31,800
Sure that these, you know, values
we populate inside.

450
00:15:31,800 --> 00:15:34,200
Each of our transactions in our list.

451
00:15:34,200 --> 00:15:37,000
Well. In order to make sure
that these are strings, we can.

452
00:15:37,000 --> 00:15:38,366
Force it this way by.

453
00:15:38,366 --> 00:15:39,900
Putting that into. The.

454
00:15:39,900 --> 00:15:41,266
String function.

455
00:15:41,266 --> 00:15:44,400
Str, which will take us input exactly
these.

456
00:15:44,400 --> 00:15:45,866
Elements, meaning this.

457
00:15:45,866 --> 00:15:47,666
Product. Of the transactions.

458
00:15:47,666 --> 00:15:47,933
All right.

459
00:15:47,933 --> 00:15:49,333
So now we make sure.

460
00:15:49,333 --> 00:15:50,766
That the products are.

461
00:15:50,766 --> 00:15:53,300
Strings you know in actually. Quotes.

462
00:15:53,300 --> 00:15:54,100
And this.

463
00:15:54,100 --> 00:15:58,466
Will give exactly what the primary model
is expecting as.

464
00:15:58,466 --> 00:16:00,366
The format of its inputs.

465
00:16:00,366 --> 00:16:01,800
Good. Perfect.

466
00:16:01,800 --> 00:16:04,200
So now we're done with data preprocessing.

467
00:16:04,200 --> 00:16:06,566
As I told you,
it's a. Bit different than before.

468
00:16:06,566 --> 00:16:07,400
But now you know.

469
00:16:07,400 --> 00:16:10,366
That for the primary model,
you have to just create a.

470
00:16:10,366 --> 00:16:12,866
List of transactions
and make sure. That all.

471
00:16:12,866 --> 00:16:15,400
The elements
in each of your transactions are.

472
00:16:15,400 --> 00:16:16,500
Strings.

473
00:16:16,500 --> 00:16:17,133
Perfect.

474
00:16:17,133 --> 00:16:20,100
So now let's execute this cell
because we're ready.

475
00:16:20,100 --> 00:16:21,900
Let's hope we didn't make any mistake.

476
00:16:21,900 --> 00:16:25,033
Play running this cell and all.

477
00:16:25,033 --> 00:16:25,933
Good. Perfect.

478
00:16:25,933 --> 00:16:27,666
It just ran successfully.

479
00:16:27,666 --> 00:16:29,400
Now we have. This list of.

480
00:16:29,400 --> 00:16:32,966
Transactions containing
all the same transactions as in.

481
00:16:32,966 --> 00:16:33,966
This data set.

482
00:16:33,966 --> 00:16:36,900
But in the format. Of a list. Good.

483
00:16:36,900 --> 00:16:38,366
So now we're.
Going to take a little break.

484
00:16:38,366 --> 00:16:40,033
Because then comes the. Real.

485
00:16:40,033 --> 00:16:41,400
Big. Important step.

486
00:16:41,400 --> 00:16:42,500
Training the a.

487
00:16:42,500 --> 00:16:45,466
Priori model on the data set.

488
00:16:45,466 --> 00:16:48,733
And to do this
we'll call this a primary function

489
00:16:49,033 --> 00:16:52,200
that will take as input exactly this list.

490
00:16:52,200 --> 00:16:54,266
Of transactions that is now. Correctly.

491
00:16:54,266 --> 00:16:59,300
Populated in the right format to train
the primary model on the data. Set.

492
00:17:00,000 --> 00:17:02,333
So take a little break
and as soon as you're ready.

493
00:17:02,333 --> 00:17:04,833
Let's implement that. Next step together.

494
00:17:04,833 --> 00:17:06,766
And until then, enjoy machine learning.
