1
00:00:00,090 --> 00:00:02,346
You now know a bunch about machine learning.

2
00:00:02,390 --> 00:00:03,635
In this video, I like to

3
00:00:03,635 --> 00:00:05,448
teach you a programing language,

4
00:00:05,470 --> 00:00:06,718
Octave, in which you'll be

5
00:00:06,760 --> 00:00:08,878
able to very quickly implement

6
00:00:08,890 --> 00:00:10,259
the the learning algorithms we've

7
00:00:10,259 --> 00:00:11,770
seen already, and the learning

8
00:00:11,770 --> 00:00:14,872
algorithms we'll see later in this course.

9
00:00:14,900 --> 00:00:16,381
In the past, I've tried to teach machine learning

10
00:00:16,381 --> 00:00:19,497
using a large variety of different programming languages

11
00:00:19,500 --> 00:00:22,046
including C++ Java,

12
00:00:22,825 --> 00:00:25,379
Python, NumPy, and also

13
00:00:25,379 --> 00:00:27,128
Octave, and what I

14
00:00:27,160 --> 00:00:28,783
found was that students were able

15
00:00:28,790 --> 00:00:30,535
to learn the most

16
00:00:30,570 --> 00:00:32,497
productively learn the most quickly

17
00:00:32,497 --> 00:00:33,780
and prototype your algorithms most

18
00:00:33,780 --> 00:00:35,569
quickly using a relatively

19
00:00:35,569 --> 00:00:38,262
high level language like octave.

20
00:00:38,290 --> 00:00:39,798
In fact, what I often

21
00:00:39,798 --> 00:00:41,516
see in Silicon Valley is

22
00:00:41,520 --> 00:00:43,655
that if even if you need to build.

23
00:00:43,655 --> 00:00:44,714
If you want to build a large

24
00:00:44,740 --> 00:00:46,548
scale deployment of a learning

25
00:00:46,610 --> 00:00:48,242
algorithm, what people will often do

26
00:00:48,242 --> 00:00:50,637
is prototype and the language is Octave.

27
00:00:50,660 --> 00:00:52,200
Which is a great prototyping language.

28
00:00:52,210 --> 00:00:55,264
So you can sort of get your learning algorithms working quickly.

29
00:00:55,270 --> 00:00:56,629
And then only if you need

30
00:00:56,629 --> 00:00:58,459
to a very large scale deployment of it.

31
00:00:58,480 --> 00:01:00,362
Only then spend your time

32
00:01:00,362 --> 00:01:03,059
re-implementing the algorithm

33
00:01:03,059 --> 00:01:05,150
to C++ Java or some of the language like that.

34
00:01:05,160 --> 00:01:06,273
Because all the lessons we've learned is

35
00:01:06,300 --> 00:01:08,679
that a time or develop a time.

36
00:01:08,710 --> 00:01:09,848
That is your time.

37
00:01:09,870 --> 00:01:13,309
The machine learning's time is incredibly valuable.

38
00:01:13,320 --> 00:01:15,101
And if you can

39
00:01:15,101 --> 00:01:17,898
get your learning algorithms to work more quickly in Octave.

40
00:01:17,898 --> 00:01:18,932
Then overall you have a

41
00:01:18,932 --> 00:01:20,697
huge time savings by first

42
00:01:20,720 --> 00:01:22,143
developing the algorithms in

43
00:01:22,150 --> 00:01:23,971
Octave, and then implementing and

44
00:01:23,971 --> 00:01:28,145
maybe C++ Java, only after we have the ideas working.

45
00:01:28,160 --> 00:01:30,238
The most common prototyping language I

46
00:01:30,238 --> 00:01:31,538
see people use for machine

47
00:01:31,560 --> 00:01:34,058
learning are: Octave, MATLAB,

48
00:01:34,070 --> 00:01:37,230
Python, NumPy, and R.

49
00:01:38,150 --> 00:01:40,032
Octave is nice because open sourced.

50
00:01:40,032 --> 00:01:42,660
And MATLAB works well

51
00:01:42,670 --> 00:01:44,656
too, but it is expensive for

52
00:01:44,656 --> 00:01:45,956
to many people.

53
00:01:45,960 --> 00:01:47,972
But if you have access to a copy of MATLAB.

54
00:01:47,988 --> 00:01:50,095
You can also use MATLAB with this class.

55
00:01:50,110 --> 00:01:52,037
If you know Python, NumPy,

56
00:01:52,037 --> 00:01:54,853
or if you know R. I do see some people use it.

57
00:01:54,870 --> 00:01:56,353
But, what I see is

58
00:01:56,360 --> 00:01:57,739
that people usually end up

59
00:01:57,760 --> 00:02:00,041
developing somewhat more slowly, and

60
00:02:00,050 --> 00:02:02,121
you know, these languages.

61
00:02:02,121 --> 00:02:04,048
Because the Python, NumPy syntax

62
00:02:04,048 --> 00:02:08,391
is just slightly clunkier than the Octave syntax.

63
00:02:08,410 --> 00:02:09,704
And so because of that, and

64
00:02:09,704 --> 00:02:11,372
because we are releasing starter

65
00:02:11,380 --> 00:02:13,039
code in Octave.

66
00:02:13,039 --> 00:02:14,363
I strongly recommend that you

67
00:02:14,363 --> 00:02:18,321
not try to do the following exercises in this class in NumPy and R.

68
00:02:18,330 --> 00:02:19,805
But that I do recommend that

69
00:02:19,805 --> 00:02:21,498
you instead do the programming exercises

70
00:02:21,520 --> 00:02:24,292
for this class in octave instead.

71
00:02:24,330 --> 00:02:25,428
What I'm going to do in

72
00:02:25,428 --> 00:02:26,708
this video is go through

73
00:02:26,708 --> 00:02:28,667
a list of commands very,

74
00:02:28,667 --> 00:02:29,879
very quickly, and its goal

75
00:02:29,879 --> 00:02:31,073
is to quickly show you the

76
00:02:31,080 --> 00:02:34,807
range of commands and the range of things you can do in Octave.

77
00:02:34,807 --> 00:02:36,493
The course website will have

78
00:02:36,520 --> 00:02:38,965
a transcript of everything I

79
00:02:38,965 --> 00:02:42,095
do, and so after

80
00:02:42,095 --> 00:02:43,185
watching this video you

81
00:02:43,185 --> 00:02:44,905
can refer to the transcript

82
00:02:44,905 --> 00:02:46,635
posted on the course website

83
00:02:46,635 --> 00:02:48,247
when you want find a command.

84
00:02:48,247 --> 00:02:50,226
Concretely, what I recommend

85
00:02:50,226 --> 00:02:53,225
you do is first watch the tutorial videos.

86
00:02:53,230 --> 00:02:55,118
And after watching to the

87
00:02:55,120 --> 00:02:58,728
end, then install Octave on your computer.

88
00:02:58,728 --> 00:02:59,738
And finally, it goes to

89
00:02:59,738 --> 00:03:01,769
the course website, download the transcripts

90
00:03:01,770 --> 00:03:02,983
of the things you see in the

91
00:03:02,983 --> 00:03:04,915
session, and type in

92
00:03:04,930 --> 00:03:07,162
whatever commands seem interesting

93
00:03:07,200 --> 00:03:09,132
to you into Octave, so that it's

94
00:03:09,132 --> 00:03:10,602
running on your own computer, so

95
00:03:10,602 --> 00:03:12,962
you can see it run for yourself.

96
00:03:12,970 --> 00:03:15,535
And with that let's get started.

97
00:03:15,920 --> 00:03:19,363
Here's my Windows desktop, and I'm going to start up Octave.

98
00:03:19,370 --> 00:03:20,977
And I'm now in Octave.

99
00:03:20,977 --> 00:03:22,522
And that's my Octave prompt.

100
00:03:22,522 --> 00:03:24,475
Let me first show the elementary

101
00:03:24,475 --> 00:03:27,291
operations you can do in Octave.

102
00:03:27,330 --> 00:03:28,505
So you type in 5 + 6.

103
00:03:28,505 --> 00:03:30,493
That gives you the answer of 11.

104
00:03:30,493 --> 00:03:31,516
3 - 2.

105
00:03:31,540 --> 00:03:33,710
5 x 8, 1/2, 2^6

106
00:03:35,733 --> 00:03:37,747
is 64.

107
00:03:37,810 --> 00:03:42,361
So those are the elementary math operations.

108
00:03:42,390 --> 00:03:44,495
You can also do logical operations.

109
00:03:44,550 --> 00:03:45,929
So one equals two.

110
00:03:45,929 --> 00:03:47,722
This evaluates to false.

111
00:03:47,722 --> 00:03:51,658
The percent command here means a comment.

112
00:03:51,658 --> 00:03:53,861
So, one equals two, evaluates to false.

113
00:03:53,861 --> 00:03:55,622
Which is represents by zero.

114
00:03:55,650 --> 00:03:58,028
One not equals to two.

115
00:03:58,028 --> 00:03:59,312
This is true.

116
00:03:59,312 --> 00:04:00,718
So that returns one.

117
00:04:00,718 --> 00:04:02,146
Note that a not equal sign

118
00:04:02,146 --> 00:04:05,478
is this tilde equals symbol.

119
00:04:05,550 --> 00:04:07,336
And not bang equals.

120
00:04:07,336 --> 00:04:09,267
Which is what some other

121
00:04:09,267 --> 00:04:10,878
programming languages use.

122
00:04:10,910 --> 00:04:13,616
Lets see logical operations one and zero

123
00:04:13,616 --> 00:04:15,545
use a double ampersand sign to

124
00:04:15,545 --> 00:04:17,340
the logical AND.

125
00:04:18,120 --> 00:04:20,188
And that evaluates false.

126
00:04:20,188 --> 00:04:23,886
One or zero is the OR operation.

127
00:04:23,900 --> 00:04:25,736
And that evaluates to true.

128
00:04:25,736 --> 00:04:27,131
And I can XOR one and

129
00:04:27,131 --> 00:04:30,333
zero, and that evaluates to one.

130
00:04:30,333 --> 00:04:32,928
This thing over on the left, this Octave 324.x

131
00:04:32,928 --> 00:04:35,683
equals 11, this is the default Octave prompt.

132
00:04:35,700 --> 00:04:37,513
It shows the, what, the

133
00:04:37,520 --> 00:04:39,150
version in Octave and so on.

134
00:04:39,150 --> 00:04:40,423
If you don't want that prompt,

135
00:04:40,450 --> 00:04:43,025
there's a somewhat cryptic command PF

136
00:04:43,025 --> 00:04:44,670
quote, greater than, greater

137
00:04:44,670 --> 00:04:46,602
than and so on,

138
00:04:46,602 --> 00:04:48,800
that you can use to change the prompt.

139
00:04:48,810 --> 00:04:51,272
And I guess this quote a string in the middle.

140
00:04:51,272 --> 00:04:53,362
Your quote, greater than, greater than, space.

141
00:04:53,400 --> 00:04:55,592
That's what I prefer my Octave prompt to look like.

142
00:04:55,592 --> 00:04:57,722
So if I hit enter.

143
00:04:57,920 --> 00:04:59,763
Oops, excuse me.

144
00:04:59,763 --> 00:05:00,786
Like so.

145
00:05:00,786 --> 00:05:02,622
PS1 like so.

146
00:05:02,622 --> 00:05:05,420
Now my Octave prompt has changed to the greater than, greater than sign.Which,

147
00:05:05,500 --> 00:05:09,263
you know, looks quite a bit better.

148
00:05:09,710 --> 00:05:12,384
Next let's talk about Octave variables.

149
00:05:12,384 --> 00:05:13,865
I can take the variable

150
00:05:13,865 --> 00:05:16,165
A and assign it to 3.

151
00:05:16,165 --> 00:05:18,421
And hit enter.

152
00:05:18,440 --> 00:05:20,043
And now A is equal to 3.

153
00:05:20,070 --> 00:05:22,861
You want to assign a variable, but you don't want to print out the result.

154
00:05:22,861 --> 00:05:26,758
If you put a semicolon, the semicolon

155
00:05:26,920 --> 00:05:30,824
suppresses the print output.

156
00:05:30,824 --> 00:05:33,160
So to do that, enter, it doesn't print anything.

157
00:05:33,160 --> 00:05:35,399
Whereas A equals 3.

158
00:05:35,420 --> 00:05:36,719
mix it, print it out,

159
00:05:36,719 --> 00:05:39,845
where A equals, 3 semicolon doesn't print anything.

160
00:05:39,850 --> 00:05:41,845
I can do string assignment.

161
00:05:41,845 --> 00:05:43,473
B equals hi

162
00:05:43,520 --> 00:05:45,047
Now if I just

163
00:05:45,047 --> 00:05:46,072
enter B it prints out the

164
00:05:46,072 --> 00:05:48,338
variable B. So B is the string hi

165
00:05:48,370 --> 00:05:51,118
C equals 3 greater than colon 1.

166
00:05:51,130 --> 00:05:54,538
So, now C evaluates the true.

167
00:05:55,710 --> 00:05:57,999
If you want to print

168
00:05:58,030 --> 00:06:00,832
out or display a variable, here's how you go about it.

169
00:06:00,832 --> 00:06:03,725
Let me set A equals Pi.

170
00:06:03,760 --> 00:06:04,985
And if I want to print

171
00:06:04,985 --> 00:06:08,545
A I can just type A like so, and it will print it out.

172
00:06:08,545 --> 00:06:10,344
For more complex printing there is

173
00:06:10,344 --> 00:06:13,674
also the DISP command which stands for Display.

174
00:06:13,710 --> 00:06:15,858
Display A just prints out A like so.

175
00:06:15,890 --> 00:06:18,337
You can also display strings

176
00:06:18,350 --> 00:06:21,392
so: DISP, sprintf, two

177
00:06:21,460 --> 00:06:24,990
decimals, percent 0.2,

178
00:06:25,260 --> 00:06:28,273
F, comma, A. Like so.

179
00:06:28,273 --> 00:06:29,863
And this will print out the string.

180
00:06:29,863 --> 00:06:31,722
Two decimals, colon, 3.14.

181
00:06:31,722 --> 00:06:33,651
This is kind of

182
00:06:33,670 --> 00:06:35,993
an old style C syntax.

183
00:06:35,993 --> 00:06:37,404
For those of you that

184
00:06:37,420 --> 00:06:39,073
have programmed C before, this is

185
00:06:39,073 --> 00:06:41,378
essentially the syntax you use to print screen.

186
00:06:41,380 --> 00:06:44,498
So the Sprintf generates a

187
00:06:44,510 --> 00:06:46,021
string that is less

188
00:06:46,021 --> 00:06:48,274
than the 2 decimals, 3.1 plus string.

189
00:06:48,290 --> 00:06:50,644
This percent 0.2 F means

190
00:06:50,644 --> 00:06:52,475
substitute A into here,

191
00:06:52,475 --> 00:06:55,926
showing the two digits after the decimal points.

192
00:06:55,926 --> 00:06:58,104
And DISP takes the string

193
00:06:58,130 --> 00:07:00,691
DISP generates it by the Sprintf command.

194
00:07:00,691 --> 00:07:01,683
Sprintf.

195
00:07:01,683 --> 00:07:03,091
The Sprintf command.

196
00:07:03,091 --> 00:07:05,835
And DISP actually displays the string.

197
00:07:05,870 --> 00:07:07,020
And to show you another

198
00:07:07,020 --> 00:07:11,360
example, Sprintf six decimals

199
00:07:11,361 --> 00:07:14,551
percent 0.6 F comma A.

200
00:07:14,930 --> 00:07:17,075
And, this should print Pi

201
00:07:17,090 --> 00:07:21,100
with six decimal places.

202
00:07:22,060 --> 00:07:25,728
Finally, I was saying, a like so, looks like this. There

203
00:07:25,740 --> 00:07:28,633
are useful shortcuts that type type formats long.

204
00:07:28,633 --> 00:07:31,759
It causes strings by default.

205
00:07:31,760 --> 00:07:33,748
Be displayed to a lot more decimal places.

206
00:07:33,748 --> 00:07:35,593
And format short is a

207
00:07:35,593 --> 00:07:37,095
command that restores the default

208
00:07:37,120 --> 00:07:40,113
of just printing a small number of digits.

209
00:07:40,600 --> 00:07:43,934
Okay, that's how you work with variables.

210
00:07:43,934 --> 00:07:47,047
Now let's look at vectors and matrices.

211
00:07:47,070 --> 00:07:49,274
Let's say I want to assign MAT A to the matrix.

212
00:07:49,280 --> 00:07:50,974
Let me show you an example: 1, 2,

213
00:07:50,980 --> 00:07:54,593
semicolon, 3, 4, semicolon, 5, 6.

214
00:07:54,600 --> 00:07:56,235
This generates a three by

215
00:07:56,240 --> 00:07:58,572
two matrix A whose first

216
00:07:58,580 --> 00:07:59,818
row is 1, 2. Second row

217
00:07:59,820 --> 00:08:02,030
3, 4. Third row is 5, 6.

218
00:08:02,030 --> 00:08:04,385
What the semicolon does is

219
00:08:04,390 --> 00:08:05,818
essentially say, go to

220
00:08:05,820 --> 00:08:07,915
the next row of the matrix.

221
00:08:07,915 --> 00:08:09,016
There are other ways to type this in.

222
00:08:09,016 --> 00:08:11,536
Type A 1, 2 semicolon

223
00:08:11,536 --> 00:08:15,046
3, 4, semicolon, 5, 6, like so.

224
00:08:15,046 --> 00:08:17,038
And that's another equivalent way of

225
00:08:17,038 --> 00:08:18,576
assigning A to be

226
00:08:18,576 --> 00:08:22,183
the values of this three by two matrix.

227
00:08:22,200 --> 00:08:23,568
Similarly you can assign vectors.

228
00:08:23,568 --> 00:08:25,532
So V equals 1, 2, 3.

229
00:08:25,560 --> 00:08:27,359
This is actually a row vector.

230
00:08:27,359 --> 00:08:29,915
Or this is a 3 by 1 vector.

231
00:08:29,940 --> 00:08:32,016
Where that is a fat Y vector,

232
00:08:32,030 --> 00:08:34,375
excuse me, not, this is

233
00:08:34,380 --> 00:08:37,998
a 1 by 3 matrix, right.

234
00:08:37,998 --> 00:08:39,256
Not 3   by 1.

235
00:08:39,256 --> 00:08:41,015
If I want to assign

236
00:08:41,015 --> 00:08:43,975
this to a column vector,

237
00:08:43,975 --> 00:08:48,778
what I would do instead is do v 1;2;3.

238
00:08:48,830 --> 00:08:50,030
And this will give me a 3 by 1.

239
00:08:50,100 --> 00:08:51,797
There's a 1 by 3 vector.

240
00:08:51,797 --> 00:08:55,892
So this will be a column vector.

241
00:08:56,250 --> 00:08:57,968
Here's some more useful notation.

242
00:08:57,968 --> 00:09:02,343
V equals 1: 0.1: 2.

243
00:09:02,343 --> 00:09:03,598
What this does is

244
00:09:03,620 --> 00:09:05,716
it sets V to the bunch

245
00:09:05,716 --> 00:09:08,714
of elements that start from 1.

246
00:09:08,714 --> 00:09:10,392
And increments and steps

247
00:09:10,410 --> 00:09:13,657
of 0.1 until you get up to 2.

248
00:09:13,660 --> 00:09:19,168
So if I do this, V is going to be this, you know, row vector.

249
00:09:19,168 --> 00:09:23,022
This is what one by eleven matrix really.

250
00:09:23,022 --> 00:09:23,739
That's 1, 1.1, 1.2, 1.3 and

251
00:09:23,739 --> 00:09:26,921
so on until we

252
00:09:27,630 --> 00:09:30,141
get up to two.

253
00:09:31,440 --> 00:09:33,269
Now, and I can also

254
00:09:33,269 --> 00:09:35,049
set V equals one colon six,

255
00:09:35,060 --> 00:09:38,270
and that sets V to be these numbers.

256
00:09:38,270 --> 00:09:41,291
1 through 6, okay.

257
00:09:41,620 --> 00:09:44,254
Now here are some other ways to generate matrices.

258
00:09:44,254 --> 00:09:47,426
Ones 2.3 is a command

259
00:09:47,426 --> 00:09:49,134
that generates a matrix that

260
00:09:49,140 --> 00:09:50,790
is a two by three matrix

261
00:09:50,790 --> 00:09:52,712
that is the matrix of all ones.

262
00:09:52,712 --> 00:09:53,991
So if I set that c2

263
00:09:54,000 --> 00:09:56,845
times ones two by

264
00:09:56,845 --> 00:09:59,798
three this generates a

265
00:09:59,798 --> 00:10:03,061
two by three matrix that is all two's.

266
00:10:03,080 --> 00:10:04,258
You can think of this as a

267
00:10:04,258 --> 00:10:05,513
shorter way of writing this and

268
00:10:05,550 --> 00:10:06,943
c2,2,2's and you can

269
00:10:06,943 --> 00:10:10,951
call them 2,2,2, which would also give you the same result.

270
00:10:11,450 --> 00:10:13,910
Let's say W equals one's, one

271
00:10:13,920 --> 00:10:15,485
by three, so this is

272
00:10:15,485 --> 00:10:17,937
going to be a row vector

273
00:10:17,940 --> 00:10:20,998
or a row of

274
00:10:20,998 --> 00:10:23,853
three one's and similarly

275
00:10:23,853 --> 00:10:25,463
you can also say w equals

276
00:10:25,463 --> 00:10:27,469
zeroes, one by

277
00:10:27,469 --> 00:10:30,209
three, and this generates a matrix.

278
00:10:30,220 --> 00:10:34,732
A one by three matrix of all zeros.

279
00:10:34,732 --> 00:10:36,910
Just a couple more ways to generate matrices .

280
00:10:36,930 --> 00:10:39,175
If I do W equals

281
00:10:39,175 --> 00:10:41,512
Rand one by three,

282
00:10:41,520 --> 00:10:43,050
this gives me a one

283
00:10:43,050 --> 00:10:45,370
by three matrix of all random numbers.

284
00:10:45,372 --> 00:10:47,118
If I do Rand

285
00:10:47,215 --> 00:10:49,008
three by three.

286
00:10:49,050 --> 00:10:50,417
This gives me a three by

287
00:10:50,417 --> 00:10:51,918
three matrix of all

288
00:10:51,930 --> 00:10:54,009
random numbers drawn from the

289
00:10:54,009 --> 00:10:55,830
uniform distribution between zero and one.

290
00:10:55,830 --> 00:10:56,937
So every time I do

291
00:10:56,937 --> 00:10:58,608
this, I get a different

292
00:10:58,608 --> 00:11:00,510
set of random numbers drawn

293
00:11:00,540 --> 00:11:02,573
uniformly between zero and one.

294
00:11:02,573 --> 00:11:03,718
For those of you that

295
00:11:03,718 --> 00:11:05,375
know what a Gaussian random variable

296
00:11:05,410 --> 00:11:06,275
is or for those of you that

297
00:11:06,275 --> 00:11:07,659
know what a normal random variable

298
00:11:07,660 --> 00:11:09,112
is, you can also set W

299
00:11:09,112 --> 00:11:11,956
equals Rand N, one by three.

300
00:11:11,990 --> 00:11:13,565
And so these are going

301
00:11:13,570 --> 00:11:15,435
to be three values drawn from

302
00:11:15,435 --> 00:11:17,798
a Gaussian distribution with mean

303
00:11:17,798 --> 00:11:19,266
zero and variance or

304
00:11:19,266 --> 00:11:21,642
standard deviation equal to one.

305
00:11:21,642 --> 00:11:23,148
And you can set more complex

306
00:11:23,150 --> 00:11:24,698
things like W equals minus

307
00:11:24,698 --> 00:11:26,194
six, plus the square root

308
00:11:26,210 --> 00:11:28,656
ten, times, lets say

309
00:11:28,660 --> 00:11:31,978
Rand N, one by ten thousand.

310
00:11:31,978 --> 00:11:33,106
And I'm going to put a semicolon at

311
00:11:33,106 --> 00:11:35,623
the end because I don't really want this printed out.

312
00:11:35,623 --> 00:11:37,599
This is going to be a what?

313
00:11:37,599 --> 00:11:38,905
Well, it's going to

314
00:11:38,910 --> 00:11:40,582
be a vector of, with

315
00:11:40,610 --> 00:11:44,481
a hundred thousand, excuse me, ten thousand elements.

316
00:11:44,490 --> 00:11:47,596
So, well, actually, you know what?

317
00:11:47,596 --> 00:11:48,373
Let's print it out.

318
00:11:48,373 --> 00:11:51,570
So this will generate a matrix like this.

319
00:11:51,570 --> 00:11:52,408
Right?

320
00:11:52,408 --> 00:11:53,978
With 10,000 elements.

321
00:11:53,978 --> 00:11:55,835
So that's what W is.

322
00:11:55,835 --> 00:11:57,392
And if I now

323
00:11:57,392 --> 00:11:59,442
plot a histogram of W

324
00:11:59,442 --> 00:12:01,818
with a hist command, I can

325
00:12:01,820 --> 00:12:04,752
now. And Octave's print hist

326
00:12:04,752 --> 00:12:06,130
command, you know, takes a

327
00:12:06,130 --> 00:12:07,297
couple seconds to bring this up,

328
00:12:07,297 --> 00:12:08,965
but this is a histogram of

329
00:12:08,970 --> 00:12:10,646
my random variable for W.

330
00:12:10,650 --> 00:12:12,732
There was minus 6 plus zero

331
00:12:12,732 --> 00:12:15,537
ten times this Gaussian random variable.

332
00:12:15,537 --> 00:12:17,537
And I can plot a histogram with

333
00:12:17,560 --> 00:12:21,032
more buckets, with more bins, with say, 50 bins.

334
00:12:21,032 --> 00:12:22,578
And this is my

335
00:12:22,578 --> 00:12:25,735
histogram of a Gaussian with mean minus 6.

336
00:12:25,735 --> 00:12:27,285
Because I have a minus

337
00:12:27,285 --> 00:12:29,208
6 there plus square root 10 times this.

338
00:12:29,230 --> 00:12:32,952
So the variance of

339
00:12:32,952 --> 00:12:34,961
this Gaussian random variable

340
00:12:34,961 --> 00:12:36,696
is 10 on the standard deviation is

341
00:12:36,700 --> 00:12:38,935
square root of 10, which is about what?

342
00:12:38,950 --> 00:12:41,063
Three point one.

343
00:12:41,780 --> 00:12:43,857
Finally, one special command

344
00:12:43,857 --> 00:12:46,208
for generator matrix, which is the I command.

345
00:12:46,208 --> 00:12:48,394
So I stands for this

346
00:12:48,394 --> 00:12:51,028
is maybe a pun on the word identity.

347
00:12:51,050 --> 00:12:52,650
It's server set eye 4.

348
00:12:52,720 --> 00:12:56,004
This is the 4 by 4 identity matrix.

349
00:12:56,004 --> 00:12:57,681
So I equals eye 4.

350
00:12:57,681 --> 00:13:00,458
This gives me a 4 by 4 identity matrix.

351
00:13:00,458 --> 00:13:04,475
And I equals eye 5,  eye 6.

352
00:13:04,475 --> 00:13:05,611
That gives me a 6 by

353
00:13:05,611 --> 00:13:08,089
6 identity matrix, i3

354
00:13:08,120 --> 00:13:09,134
is the 3 by 3 identity matrix.

355
00:13:09,134 --> 00:13:12,064
Lastly, to

356
00:13:12,064 --> 00:13:14,263
wrap up this video, there's one more useful command.

357
00:13:14,280 --> 00:13:15,479
Which is the help command.

358
00:13:15,479 --> 00:13:17,454
So you can type help i and

359
00:13:17,454 --> 00:13:21,181
this brings up the help function for the identity matrix.

360
00:13:21,190 --> 00:13:22,803
Hit Q to quit.

361
00:13:22,803 --> 00:13:25,375
And you can also type help rand.

362
00:13:25,380 --> 00:13:27,793
Brings up documentation for the rand or the

363
00:13:27,793 --> 00:13:29,734
random number generation function.

364
00:13:29,734 --> 00:13:31,898
Or even help help, which

365
00:13:31,900 --> 00:13:35,615
shows you, you know help on the help function.

366
00:13:36,455 --> 00:13:39,022
So, those are the

367
00:13:39,022 --> 00:13:41,612
basic operations in Octave.

368
00:13:41,612 --> 00:13:42,699
And with this you should be

369
00:13:42,699 --> 00:13:47,131
able to generate a few matrices, multiply, add things.

370
00:13:47,131 --> 00:13:50,553
And use the basic operations in Octave.

371
00:13:50,560 --> 00:13:51,893
In the next video, I'd like

372
00:13:51,920 --> 00:13:53,818
to start talking about more

373
00:13:53,818 --> 00:13:55,700
sophisticated commands and how

374
00:13:55,750 --> 00:13:59,180
to use data around and start to process data in Octave.