1 00:00:02,350 --> 00:00:08,960 This is what is known in software engineering as test driven development, or TTD in test driven development. 2 00:00:08,980 --> 00:00:13,740 We start by running tests because, number one, your code should always have tests. 3 00:00:14,140 --> 00:00:20,020 And number two, running the test first forces you to think about how you expect your API to work, 4 00:00:20,320 --> 00:00:26,010 and it allows you to make sensible design decisions without being distracted by implementation details. 5 00:00:26,740 --> 00:00:31,360 So you see how this applies to every machine learning algorithm, not just linear regression. 6 00:00:35,580 --> 00:00:40,980 And a lot of my older coding videos, I unfortunately decided to just write the code straight through 7 00:00:41,550 --> 00:00:44,220 this helps some people, but it's not ideal for others. 8 00:00:44,730 --> 00:00:47,880 In addition, the coding videos will appear without warning. 9 00:00:48,610 --> 00:00:51,130 Generally, the lectures appear in the same pattern. 10 00:00:51,610 --> 00:00:54,500 First we do theory and then we do the corresponding code. 11 00:00:54,840 --> 00:00:56,700 It pretty much always works this way. 12 00:00:57,810 --> 00:01:02,970 So after the theory lecture, you can be pretty sure we are going to implement that theory in code soon, 13 00:01:03,210 --> 00:01:09,510 unless it's so simple that it can be an exercise now to avoid treating everyone like they're in kindergarten. 14 00:01:09,750 --> 00:01:12,630 I don't have videos telling you that it's time to code. 15 00:01:13,200 --> 00:01:15,690 You're all proactive and independent adults. 16 00:01:15,900 --> 00:01:21,240 So when you learn some theory, you already know that it's time to try to put that into code by yourself. 17 00:01:25,550 --> 00:01:31,400 Sometimes you might get stuck and have no idea what's a writing code or what the data set is, in that 18 00:01:31,400 --> 00:01:35,690 case, it's fine to watch the next coding lecture, but don't watch it all the way through. 19 00:01:36,290 --> 00:01:41,480 Watch just enough so that you know what the general idea is and what we're trying to accomplish and 20 00:01:41,480 --> 00:01:43,300 then try to finish the rest of yourself. 21 00:01:47,970 --> 00:01:52,710 Now, let's say you've watched just enough so that you know what we're trying to do, you know the theory, 22 00:01:52,740 --> 00:01:58,680 but you still can't complete the code at this point, you may want to revisit the lecture theory to 23 00:01:58,680 --> 00:02:00,480 make sure you really understood each part. 24 00:02:01,230 --> 00:02:05,370 Failing that, you may want to try and get clarification on the discussion board. 25 00:02:05,880 --> 00:02:09,210 Failing that, you may want to look at external resources as well. 26 00:02:09,990 --> 00:02:14,910 Sometimes you might be missing some important computer science background that I've assumed everyone 27 00:02:14,910 --> 00:02:15,720 already knows. 28 00:02:16,290 --> 00:02:19,430 That's perfectly fine because everyone has a different background. 29 00:02:20,070 --> 00:02:22,410 It's highly encouraged to do your own research. 30 00:02:22,410 --> 00:02:24,510 And this is also perfectly normal. 31 00:02:24,900 --> 00:02:26,010 It's not a bad thing. 32 00:02:26,010 --> 00:02:26,820 It's a good thing. 33 00:02:27,420 --> 00:02:32,180 Ask any university student if this is what they do and they'll tell you that they do it all the time. 34 00:02:32,940 --> 00:02:36,150 This builds a level of mental toughness and initiative taking. 35 00:02:36,390 --> 00:02:37,200 That's important. 36 00:02:37,200 --> 00:02:42,450 If you want to learn this stuff, there is a lot to learn and I myself am still learning new things 37 00:02:42,450 --> 00:02:42,960 every day. 38 00:02:46,330 --> 00:02:51,430 So let's say you've watched the theory lecture, you've looked at some external resources on the theory, 39 00:02:51,640 --> 00:02:54,380 and you still don't understand how you would put it into code. 40 00:02:55,240 --> 00:02:58,080 Well, at this point, there are some great exercises you can do. 41 00:02:58,840 --> 00:03:04,270 Watch the coding lecture or look at the focus on GitHub, go through it and make sure you understand 42 00:03:04,270 --> 00:03:04,870 each line. 43 00:03:05,860 --> 00:03:10,270 The next step is without looking at the code again, try to implement it yourself. 44 00:03:10,870 --> 00:03:12,860 This by itself is a great learning tool. 45 00:03:13,510 --> 00:03:18,700 Many of the algorithms I've learned, I didn't understand them the first time I learned them by looking 46 00:03:18,700 --> 00:03:21,370 at someone else's implementation and dissecting it. 47 00:03:23,240 --> 00:03:28,250 I looked at each part and I asked myself, which parts did I get right and which parts did I get wrong? 48 00:03:29,150 --> 00:03:32,180 This helps you understand the mistakes you were making previously. 49 00:03:35,820 --> 00:03:41,970 The cool thing about implementation is it works backwards to we go from theory to code, but thinking 50 00:03:41,970 --> 00:03:45,940 in the reverse direction, the code actually helps us understand the theory better. 51 00:03:45,960 --> 00:03:51,350 Also, I always say that if you can't implement the code, then you don't understand the theory. 52 00:03:52,140 --> 00:03:57,930 So if you can learn the theory 50 percent of the way, try to implement the code, look at an existing 53 00:03:57,930 --> 00:04:00,230 implementation and then finally get it working. 54 00:04:00,480 --> 00:04:04,860 Then you can be guaranteed that you now know more than just 50 percent of the theory. 55 00:04:05,460 --> 00:04:10,350 The active coding helps you understand more of the theory than if you had just learned the theory by 56 00:04:10,350 --> 00:04:10,890 itself. 57 00:04:14,100 --> 00:04:19,470 So implementation helps you understand the theory, but it also helps you understand the intuition as 58 00:04:19,470 --> 00:04:19,820 well. 59 00:04:20,600 --> 00:04:22,290 Intuition is a tricky thing. 60 00:04:22,320 --> 00:04:25,140 In fact, I would go so far as to say it's a trap. 61 00:04:25,710 --> 00:04:30,000 In many or maybe even most cases, intuition is built on experience. 62 00:04:30,540 --> 00:04:35,910 That means driving the theory from first principles or in other words, driving the theory from facts 63 00:04:35,910 --> 00:04:38,990 you already know to be true and from implementation. 64 00:04:39,870 --> 00:04:45,210 Students often want to hear intuition from people with lots of experience, but the problem is that 65 00:04:45,210 --> 00:04:47,340 this intuition is built on hindsight. 66 00:04:51,170 --> 00:04:57,350 Intuition is often both the first thing and the last thing you want to focus on, you want to have a 67 00:04:57,350 --> 00:05:02,910 little bit of intuition at the beginning so you have some general idea of what you want to accomplish. 68 00:05:03,590 --> 00:05:08,960 Then after you learn the theory and implementation, you can reflect on the details and summarize them 69 00:05:08,960 --> 00:05:09,690 at a high level. 70 00:05:10,280 --> 00:05:13,580 This is real intuition built on hindsight and experience. 71 00:05:18,350 --> 00:05:23,690 The big mistake beginners make is that they read some intuition written by someone else who actually 72 00:05:23,690 --> 00:05:26,120 has experience when they themselves do not. 73 00:05:27,050 --> 00:05:30,140 From an outsider's perspective, it's hard to tell the difference. 74 00:05:30,710 --> 00:05:31,430 Why is that? 75 00:05:32,570 --> 00:05:37,730 Well, if the beginner can regurgitate the same high level ideas as the expert, then it would sound 76 00:05:37,730 --> 00:05:40,280 like they are also an expert when in fact they are not. 77 00:05:41,320 --> 00:05:46,570 There are unfortunately, too many courses out there that do this, they want you to feel good that 78 00:05:46,570 --> 00:05:50,670 you learn something and they want to avoid you having to do anything too difficult. 79 00:05:51,670 --> 00:05:56,550 And because you can repeat the same high level intuitions, it actually sounds like you're an expert. 80 00:05:57,070 --> 00:05:58,750 So this is why I call it a trap. 81 00:06:03,370 --> 00:06:08,710 Knowing intuition not only makes you feel like you know something, it also makes you really confident 82 00:06:08,710 --> 00:06:09,640 about this fact. 83 00:06:10,300 --> 00:06:16,060 And then because your ego wants to maintain this confidence, you'll start making crazy statements like 84 00:06:16,060 --> 00:06:21,160 the theory and implementation are not important because you already understand the intuition. 85 00:06:21,760 --> 00:06:24,130 But as you know, this is completely backwards. 86 00:06:28,150 --> 00:06:34,180 Let's summarize this lecture, since that was a lot of information to take in, remember, this lecture 87 00:06:34,180 --> 00:06:36,870 is all about how and why you should code by yourself. 88 00:06:37,420 --> 00:06:43,270 First, we talked about why this is because it ensures you are thinking about the details and filling 89 00:06:43,270 --> 00:06:46,060 in the blanks using your own pattern recognition skills. 90 00:06:46,660 --> 00:06:51,160 It ensures you don't just assume you know what's going on, but that you actually know what's going 91 00:06:51,160 --> 00:06:51,490 on. 92 00:06:52,210 --> 00:06:57,160 There's a big difference, especially if you're the type of person that likes to assume you know everything. 93 00:06:58,120 --> 00:07:02,350 Next, we talked about how we talked about how all data is the same. 94 00:07:02,350 --> 00:07:06,130 And it doesn't really matter what domain or industry your data is from. 95 00:07:06,370 --> 00:07:08,090 The same algorithms all apply. 96 00:07:08,950 --> 00:07:14,160 Next, we looked at the interfaces to the algorithms, which are also conveniently all the same. 97 00:07:14,770 --> 00:07:20,800 So we have a kind of mix and match type of scenario where we can input any kind of data into any kind 98 00:07:20,800 --> 00:07:21,280 of model. 99 00:07:22,130 --> 00:07:28,010 For example, you can use linear regression with financial data or you can use a neural network on ecological 100 00:07:28,010 --> 00:07:32,450 data, all data is the same and all models have the same interface. 101 00:07:36,310 --> 00:07:42,310 Next, we talked about a strategy for how to actually apply this, you've probably already seen that 102 00:07:42,310 --> 00:07:44,670 we always alternate between theory and code. 103 00:07:45,250 --> 00:07:48,610 First we discuss the theory and then we implement that in code. 104 00:07:49,450 --> 00:07:54,370 Sometimes you might have to go back and revisit the theory or sometimes you might have to look things 105 00:07:54,370 --> 00:07:55,230 up on the Internet. 106 00:07:55,600 --> 00:07:56,830 That's perfectly normal. 107 00:07:57,870 --> 00:08:00,970 This is pretty much the day in the life of a university student. 108 00:08:01,680 --> 00:08:06,310 We're trying to build the same level of mental toughness and initiative taking in this course. 109 00:08:06,960 --> 00:08:11,790 So the strategy is stop after you learn the theory and try to code yourself. 110 00:08:12,710 --> 00:08:17,600 Sometimes you need a little more detail, so you might want to watch a little bit of the coding lectures 111 00:08:17,600 --> 00:08:20,960 to see what's going on and then try to implement the rest of yourself. 112 00:08:21,680 --> 00:08:26,600 If you can't get it by yourself, then maybe try watching the whole thing or looking at the expected 113 00:08:26,600 --> 00:08:29,340 result and then trying to implement it from memory. 114 00:08:30,140 --> 00:08:34,850 This by itself is a great learning tool because it allows you to ask yourself, what did I get right 115 00:08:34,850 --> 00:08:36,020 and what did I get wrong? 116 00:08:36,290 --> 00:08:38,450 So you can understand the mistakes you were making. 117 00:08:42,850 --> 00:08:48,550 Finally, it's important to understand the difference between intuition, theory and implementation 118 00:08:48,670 --> 00:08:50,300 and how they reinforce each other. 119 00:08:51,040 --> 00:08:56,770 It's important to not fall into the trap of only understanding intuition written by someone else, and 120 00:08:56,770 --> 00:09:02,740 then assuming that this equates to actually understanding everything in many cases, it's not about 121 00:09:02,740 --> 00:09:06,640 waiting for someone to tell you some nice phrases that trick you into thinking. 122 00:09:06,640 --> 00:09:12,310 You understand the intuition, but rather you should seek intuition by working through the theory and 123 00:09:12,310 --> 00:09:13,330 the implementation. 124 00:09:13,900 --> 00:09:18,040 And lastly, don't forget, when it's time to code, you must code.