1 00:00:19,000 --> 00:00:24,460 Hi! Well, Class 2 of Data Mining with Weka has just started. 2 00:00:24,469 --> 00:00:27,369 Class 1 has gone by without too many hitches. 3 00:00:27,369 --> 00:00:31,710 I've enjoyed looking at your comments on the mailing list. 4 00:00:31,710 --> 00:00:35,910 Thank you for answering each other's questions, and thank you, Peter, for answering so many 5 00:00:35,910 --> 00:00:36,879 technical questions. 6 00:00:36,879 --> 00:00:43,879 I just thought I'd talk briefly about some of the common issues that have arisen. 7 00:00:45,190 --> 00:00:48,930 First of all, there is the website and how to get to the course. 8 00:00:48,930 --> 00:00:52,190 Some people were going straight to the YouTube videos, and if you just go straight to the 9 00:00:52,190 --> 00:00:54,160 YouTube videos you don't see the activities. 10 00:00:54,160 --> 00:00:57,829 You should be seeing this picture when you go to the course. 11 00:00:57,829 --> 00:01:04,290 This is the website, and you need to go and look at the course from here. 12 00:01:04,290 --> 00:01:07,280 Go to Class 1 and Class 2 and so on from here. 13 00:01:07,280 --> 00:01:11,479 This is the entry point to the course. 14 00:01:11,479 --> 00:01:16,450 Another problem for some has been installing Weka on your computer. 15 00:01:16,450 --> 00:01:22,040 I guess I should have said that since Weka is written in Java you need Java on your computer, 16 00:01:22,040 --> 00:01:26,820 or you need to install Java first, or install Java as part of Weka. 17 00:01:26,820 --> 00:01:31,150 One of the problems people were having is that they didn't have Java installed. 18 00:01:31,150 --> 00:01:35,369 Let me just show you how to test for whether you have Java installed or not. 19 00:01:35,369 --> 00:01:41,810 If you go to your Windows Start Menu -- this is just on Windows -- if you type 'cmd' you 20 00:01:41,810 --> 00:01:47,350 get a command line, or command window. 21 00:01:47,350 --> 00:01:52,119 I call this the "black screen of death," actually: we often don't like to see this. 22 00:01:52,119 --> 00:02:01,990 But if you just simply type 'java' and it comes back with this, then Java is installed 23 00:02:01,990 --> 00:02:03,409 on your computer. 24 00:02:03,409 --> 00:02:08,289 If it comes back with 'cannot find Java' or something like that, then you need to first 25 00:02:08,289 --> 00:02:13,360 of all figure out how to get Java going on your computer, and then get Weka going. 26 00:02:13,360 --> 00:02:16,950 So, I just thought I'd mention that. 27 00:02:16,950 --> 00:02:22,060 A number of people have made comments about the book. 28 00:02:22,060 --> 00:02:26,290 Someone asked if it was really necessary to do the readings; they were finding the course 29 00:02:26,290 --> 00:02:28,010 quite easy. 30 00:02:28,010 --> 00:02:31,319 The answer is that it's not really necessary to do the readings, and you're *supposed* 31 00:02:31,319 --> 00:02:32,950 to find the course easy. 32 00:02:32,950 --> 00:02:39,900 It might get a little tougher in the weeks to come, but still, it's a pretty easy course. 33 00:02:39,900 --> 00:02:44,120 The readings are there for additional background, and you certainly shouldn't feel that you 34 00:02:44,120 --> 00:02:45,510 have to do them at all. 35 00:02:45,510 --> 00:02:47,920 You can do the whole course without looking at the book. 36 00:02:47,920 --> 00:02:52,599 We're interested in ensuring that people at all sorts of different levels who start out 37 00:02:52,599 --> 00:02:56,090 this course can succeed. 38 00:02:56,090 --> 00:02:58,080 You don't have to read the book. 39 00:02:58,080 --> 00:03:01,870 Someone else asked if the second edition of the book is ok. 40 00:03:01,870 --> 00:03:06,620 This is the second edition of Data Mining with Weka: you can see the cover here. 41 00:03:06,620 --> 00:03:13,620 I kind of like this one -- it's got a chameleon hidden here amongst New Zealand fern leaves. 42 00:03:14,750 --> 00:03:23,330 The third edition, this one here, is the latest edition -- this has got a tiger hidden in 43 00:03:23,330 --> 00:03:25,870 the grass. 44 00:03:25,870 --> 00:03:28,150 The answer is: the second edition is fine. 45 00:03:28,150 --> 00:03:34,430 Either of those editions are just fine if you're looking at the readings. 46 00:03:34,430 --> 00:03:38,849 Someone else said "I hate having to read it online". 47 00:03:38,849 --> 00:03:44,000 I completely agree with you! I would love to be able to provide you with a free physical 48 00:03:44,000 --> 00:03:50,159 copy of the whole book, but unfortunately I'm not able to do that. 49 00:03:50,159 --> 00:03:51,860 The realities of publishing. 50 00:03:51,860 --> 00:03:56,769 I guess the publisher is trying to increase sales, and they're hoping that you will be 51 00:03:56,769 --> 00:04:00,750 tempted to go out and buy a copy and recommend it to your friends. 52 00:04:00,750 --> 00:04:04,549 This book makes a great Christmas present, by the way -- so you can give a copy to all 53 00:04:04,549 --> 00:04:10,420 of your friends at Christmas! We can't provide you with complete PDF file that you can take 54 00:04:10,420 --> 00:04:12,790 away, and we can't provide you with a physical copy. 55 00:04:12,790 --> 00:04:18,280 I'm really sorry about that, but it's just the way it is. 56 00:04:18,280 --> 00:04:21,599 The next thing that I wanted to mention was the irises. 57 00:04:21,599 --> 00:04:23,199 This is kind of funny I thought. 58 00:04:23,199 --> 00:04:30,199 Let me just go to Class 1 here. 59 00:04:30,939 --> 00:04:34,550 This, of course, is how you're supposed to be looking at the course. 60 00:04:34,550 --> 00:04:40,169 This is Lesson 1.3, and this is where you can watch the video from. 61 00:04:40,169 --> 00:04:46,339 Then, if you go to the activity -- that's this menu item here -- this gives you the 62 00:04:46,339 --> 00:04:47,009 activity. 63 00:04:47,009 --> 00:04:53,559 We had a little bit of a problem with this question, with these iris pictures. 64 00:04:53,559 --> 00:05:02,690 Originally, we had these a's, b's and c's permuted in different order for each of the 65 00:05:02,699 --> 00:05:05,580 possible answers. 66 00:05:05,580 --> 00:05:10,110 We were trying to make sure that you really concentrate on reading these answers and don't 67 00:05:10,110 --> 00:05:12,639 just quickly scan through them. 68 00:05:12,639 --> 00:05:20,130 We were hoist by our own petard! -- that's an English phrase that means you're injured 69 00:05:20,139 --> 00:05:25,119 by the device that you intended to use to injure others, or, in our case, confused by 70 00:05:25,119 --> 00:05:29,349 the device that we intended to confuse you. 71 00:05:29,349 --> 00:05:34,180 We screwed up, and a couple of our answers were permuted versions of the same thing. 72 00:05:34,180 --> 00:05:35,529 We've fixed that now. 73 00:05:35,529 --> 00:05:40,210 This is the current page, so they are all a, b, c, in the right order. 74 00:05:40,210 --> 00:05:46,309 We thought we should make it simpler, because it seems like even we couldn't understand 75 00:05:46,309 --> 00:05:47,330 the way we had it originally. 76 00:05:47,330 --> 00:05:51,639 I thought that was quite funny actually. 77 00:05:51,639 --> 00:05:53,930 The next thing is about the algorithms. 78 00:05:53,930 --> 00:05:57,909 People want to learn about the details of the algorithms and how they work. 79 00:05:57,909 --> 00:06:02,749 Are you going to learn about those? Is there a MOOC class that goes into the algorithms 80 00:06:02,749 --> 00:06:07,939 provided by Weka, rather than the mechanics of running it? 81 00:06:07,939 --> 00:06:12,939 The answer is "yes": you will be learning something about these algorithms. 82 00:06:12,939 --> 00:06:17,509 I've put the syllabus here; it's on the course webpage. 83 00:06:17,509 --> 00:06:20,050 You can see from this syllabus what we're going to be doing. 84 00:06:20,050 --> 00:06:26,559 We're going to be looking at, for example, the J48 algorithm for decision trees and pruning 85 00:06:26,559 --> 00:06:32,529 decision trees; the nearest neighbor algorithm for instance-based learning; and linear regression; 86 00:06:32,529 --> 00:06:33,710 classification by regression. 87 00:06:33,710 --> 00:06:40,360 We'll look at quite a few algorithms in Classes 3 and 4. 88 00:06:40,360 --> 00:06:46,330 I'm not going to tell you about the algorithms in gory detail, however: they can get quite 89 00:06:46,330 --> 00:06:47,020 tricky inside. 90 00:06:47,020 --> 00:06:52,789 What I want to do is to communicate the overall way that they work -- the idea behind the 91 00:06:52,789 --> 00:06:55,899 algorithms -- rather than the details. 92 00:06:55,899 --> 00:07:00,159 The book does give you full details of exactly how these algorithms work inside. 93 00:07:00,159 --> 00:07:03,499 We're not going to be able to cover them in that much detail in the course, but we will 94 00:07:03,499 --> 00:07:08,999 be talking about how the algorithms work and what they do. 95 00:07:08,999 --> 00:07:14,709 I forgot to say when I was talking about those irises is that someone pointed out that the 96 00:07:14,709 --> 00:07:18,580 Iris Versicolor is Quebec's floral emblem. 97 00:07:18,580 --> 00:07:21,499 Thank you very much for pointing that out! I didn't know that. 98 00:07:21,499 --> 00:07:27,600 I lived in Canada for 11 years, and I didn't know that the Iris Versicolor was Quebec's flower. 99 00:07:27,600 --> 00:07:32,249 That was very nice to learn; thank you. 100 00:07:32,249 --> 00:07:36,529 The next thing I want to talk about: someone asked about using Naive Bayes. 101 00:07:36,529 --> 00:07:42,559 How can we use the NaiveBayes classifier algorithm on a dataset, and how can we test for particular 102 00:07:42,559 --> 00:07:46,039 data, whether it fits into particular classes? 103 00:07:46,039 --> 00:07:50,439 Let me go to Weka here. 104 00:07:50,439 --> 00:07:56,189 We're going to be covering this in future lessons -- Lesson 3.3 on Naive Bayes and so 105 00:07:56,189 --> 00:07:57,689 on -- but I'll just show you. 106 00:07:57,689 --> 00:07:58,989 All of this is very easy. 107 00:07:58,989 --> 00:08:03,349 If I go to Classify, and I want to run Naive Bayes, I just need to find NaiveBayes. 108 00:08:03,349 --> 00:08:09,899 I happen to know it's in the bayes section, and I can run it here. 109 00:08:09,899 --> 00:08:10,520 Just like that. 110 00:08:10,520 --> 00:08:12,009 We've just run NaiveBayes. 111 00:08:12,009 --> 00:08:19,009 I'll be doing this more slowly and looking more at the output in Lesson 3.3. 112 00:08:19,699 --> 00:08:24,839 A natural thing to ask is if you had a particular test instance, which way would Naive Bayes 113 00:08:24,839 --> 00:08:28,699 classify it, or any other kind of classifier? 114 00:08:28,699 --> 00:08:35,699 This is the weather data we're using here, and I've created a file and called it weather.one.day.arff. 115 00:08:37,190 --> 00:08:43,120 It's a standard ARFF file, and I got it by editing the weather.nominal.arff file. 116 00:08:43,120 --> 00:08:45,300 You can see that I've just got one day here. 117 00:08:45,300 --> 00:08:49,199 I've got the same header as for the regular weather file and just one day -- but I could 118 00:08:49,199 --> 00:08:51,069 have several days if I wanted. 119 00:08:51,069 --> 00:08:56,800 I've put a question mark for the class, because I want to know what class is predicted for that. 120 00:08:56,800 --> 00:09:02,680 We'll be talking about this in Lesson 2.1 -- you're probably doing it right now -- but 121 00:09:02,680 --> 00:09:05,579 we can use a "supplied test set". 122 00:09:05,579 --> 00:09:11,700 I'm going to set that one that I created, which I called weather.one.day.arff, 123 00:09:11,700 --> 00:09:16,649 as my test set. 124 00:09:16,649 --> 00:09:20,899 I can run this and it will evaluate it on the test set. 125 00:09:20,899 --> 00:09:25,970 On the "More options..." menu -- you'll be learning about this in Lesson 4.3 -- there's 126 00:09:25,970 --> 00:09:29,160 an "Output predictions" option, here. 127 00:09:29,160 --> 00:09:40,490 If I now run it and look up here, I will find instance number 1, the actual class was "?" -- I 128 00:09:40,490 --> 00:09:46,290 showed you that, that was what was in the ARFF file -- and the predicted class is "no". 129 00:09:46,290 --> 00:09:48,860 There's some other information. 130 00:09:48,860 --> 00:09:54,329 This is how I can find out what predictions would be on new test data. 131 00:09:54,329 --> 00:09:59,830 Actually, there's nothing stopping me from setting as my test file 132 00:09:59,830 --> 00:10:02,010 the same as the training file. 133 00:10:02,019 --> 00:10:08,560 I can use weather.nominal.arff as my test file, and run it again. 134 00:10:08,560 --> 00:10:15,490 Now, I can see these are the 14 instances in the standard weather data. 135 00:10:15,490 --> 00:10:22,490 This is their actual class, this is the predicted class, predicted by, in this case, Naive Bayes. 136 00:10:23,990 --> 00:10:28,300 There's a mark in this column whenever there's an error, whenever the actual class differs 137 00:10:28,300 --> 00:10:30,269 from the predicted class. 138 00:10:30,269 --> 00:10:36,490 Again, we get that by, in the "More options..." menu, checking "Output predictions". 139 00:10:36,490 --> 00:10:40,129 We're going to talk about that in other lessons. 140 00:10:40,129 --> 00:10:44,790 I just wanted to show you that it's very easy to do these things in Weka. 141 00:10:44,790 --> 00:10:49,790 The final thing I just wanted to mention is, if you're configuring a classifier -- any 142 00:10:49,790 --> 00:10:53,529 classifier, or indeed any filter -- there are these buttons at the bottom. 143 00:10:53,529 --> 00:11:00,110 There's an "Open" and "Save" button, as well as the OK button that we normally use. 144 00:11:00,110 --> 00:11:05,990 These buttons are not about opening files in the Explorer, they're about saving configured 145 00:11:05,990 --> 00:11:07,470 classifiers. 146 00:11:07,470 --> 00:11:13,060 So you could set parameters here and save that configuration with a name and a file, 147 00:11:13,060 --> 00:11:14,519 and then open it later on. 148 00:11:14,519 --> 00:11:20,339 We don't do that in this course, so we never use these Open and Save buttons here in the 149 00:11:20,339 --> 00:11:22,680 GenericObjectEditor. 150 00:11:22,680 --> 00:11:29,680 This is the GenericObjectEditor that I get by clicking a classifier or filter. 151 00:11:29,839 --> 00:11:32,230 Just ignore the Open and Save buttons here. 152 00:11:32,230 --> 00:11:37,439 They do not open ARFF files for you. 153 00:11:37,439 --> 00:11:39,199 That's all I wanted to say. 154 00:11:39,199 --> 00:11:40,339 Carry on with Class 2. 155 00:11:40,339 --> 00:11:42,379 It's great to see so many people doing this course. 156 00:11:42,379 --> 00:11:46,400 Keep having fun, and we'll talk to you later. 157 00:11:46,400 --> 00:11:47,500 Bye for now!