1 00:00:00,450 --> 00:00:01,330 Hey there. 2 00:00:01,470 --> 00:00:02,500 This is Future Daniel. 3 00:00:02,790 --> 00:00:03,690 Why future Daniel. 4 00:00:03,780 --> 00:00:08,520 Well because I've just finished off recording the last video for this entire section. 5 00:00:08,520 --> 00:00:13,830 The psychic land section so you might see in an upcoming video I save to a list. 6 00:00:13,980 --> 00:00:15,030 What we're covering. 7 00:00:15,030 --> 00:00:17,640 Section by section an end to and workflow. 8 00:00:17,730 --> 00:00:19,040 Getting the data ready. 9 00:00:19,050 --> 00:00:24,120 Choosing the right estimated slash machine learning algorithm for our problems Fitting the model algorithm 10 00:00:24,270 --> 00:00:29,670 to the data and using it to make predictions and a few more steps and what you'll probably quickly realize 11 00:00:29,670 --> 00:00:31,770 is that this a pretty big section. 12 00:00:31,770 --> 00:00:38,370 So if I scroll through the entire notebook all the code that we're gonna cover there is a lot here. 13 00:00:38,370 --> 00:00:38,890 Right. 14 00:00:38,890 --> 00:00:39,620 Right through. 15 00:00:39,640 --> 00:00:41,330 Keep following through this is don't worry. 16 00:00:41,340 --> 00:00:48,600 This is across the next 25 30 or so videos that you're gonna go through and here's that list that I 17 00:00:48,600 --> 00:00:49,800 created right at the start right. 18 00:00:49,890 --> 00:00:56,490 I think this is an in an upcoming video and now seeing this to start out with can be pretty intimidating. 19 00:00:56,490 --> 00:01:02,040 That's what I want to sort of warn you right is that because there is so much because we are going through 20 00:01:02,040 --> 00:01:07,680 a lot of different concepts of not only machine learning but psychic learn as itself. 21 00:01:07,680 --> 00:01:12,540 I wanted to make an auxiliary resource that you can kind of come to and get a quick overview of what 22 00:01:12,540 --> 00:01:13,370 we're doing. 23 00:01:13,380 --> 00:01:20,820 So this list although it tells you what's happening it doesn't really tell you much other than just 24 00:01:20,820 --> 00:01:23,060 the headings of the different sections. 25 00:01:23,190 --> 00:01:27,240 So that's why there's resource that'll be available in the resources section there'll be some sort of 26 00:01:27,480 --> 00:01:29,100 link in a place that you can find it. 27 00:01:29,190 --> 00:01:30,210 So don't worry. 28 00:01:30,870 --> 00:01:33,690 This is what we're covering in the cyclone introduction. 29 00:01:33,810 --> 00:01:39,790 So this is kind of like a table of contents for this massive notebook that we're going to go through. 30 00:01:39,930 --> 00:01:45,490 And don't worry even ones going through the entire cyclone section you're probably not going to know 31 00:01:45,490 --> 00:01:46,380 all this off by heart. 32 00:01:46,950 --> 00:01:51,930 And if it seems like I know all this off by heart and I were going through it I'm seeming like wow I 33 00:01:51,930 --> 00:01:53,060 could never do that. 34 00:01:53,070 --> 00:01:54,460 Don't be mistaken. 35 00:01:54,630 --> 00:01:59,190 I've had a fair bit of practice with the cyclone library and that's why I'm able to go through and sort 36 00:01:59,190 --> 00:02:04,110 of figure out things and know where to look and go to the documentation and go I can see this function 37 00:02:04,110 --> 00:02:06,270 to do that this function to do that. 38 00:02:06,780 --> 00:02:12,660 So going through it for the first time remind yourself learning something new takes time and learning 39 00:02:12,660 --> 00:02:16,750 anything worthwhile definitely takes time especially machine learning. 40 00:02:16,890 --> 00:02:21,180 And so this is the framework that we're going to be using for the introduction we saw this in the previous 41 00:02:21,180 --> 00:02:26,550 video that Kino we went through step one we're going to get the data ready we're gonna find out how 42 00:02:26,550 --> 00:02:31,690 to pick a cyclone model we're going to fit model system data so we can make a prediction we're gonna 43 00:02:31,740 --> 00:02:36,570 evaluate our model see if it's worthwhile even using the model that we've got you'll see there's a couple 44 00:02:36,570 --> 00:02:41,130 of different ways to use this depending on the model we're using will improve through experimentation 45 00:02:41,130 --> 00:02:47,220 because after all machine learning is all about experimentation and the quicker you can reduce your 46 00:02:47,220 --> 00:02:51,570 time between experiments the more you're going to get done the more insights you're going to figure 47 00:02:51,570 --> 00:02:56,580 out about your data about your machine learning models and they're going to look at saving and reloading 48 00:02:56,670 --> 00:03:02,100 trained machine learning models so you don't have to go through all these steps again you can save a 49 00:03:02,100 --> 00:03:07,860 model send it to a friend send it to a colleague put it in your application and use it. 50 00:03:08,000 --> 00:03:13,220 That was a bit of a mouthful in itself but if you go through this resource you'll see that I've got 51 00:03:13,220 --> 00:03:19,190 at the top standard library imports treat this as like the mini version of this. 52 00:03:19,250 --> 00:03:24,410 So this is the big behemoth that we're about to go through in this section through all 30 or so videos 53 00:03:24,830 --> 00:03:27,790 but this is like the mini overview you can quickly refer to. 54 00:03:27,800 --> 00:03:31,150 So you're not scrolling back and forth through all the stuff that's going in here. 55 00:03:31,620 --> 00:03:37,470 So to begin with we've got some standard library imports we're going to see how we can import data sets 56 00:03:37,940 --> 00:03:43,160 to use examples in the psychic line we're going to focus on a classification and a regression problem 57 00:03:43,580 --> 00:03:46,600 using our heart disease and Boston data sets. 58 00:03:46,610 --> 00:03:48,330 Step one is getting the data ready. 59 00:03:48,380 --> 00:03:48,680 Right. 60 00:03:48,680 --> 00:03:54,170 So each of these sections refers to a little step in this workflow get the data ready you'll see the 61 00:03:54,170 --> 00:03:59,930 functions that we use there you'll see Section two is how we pick a model to how we choose a machine 62 00:03:59,930 --> 00:04:03,770 learning model for what problem we're working on this is one of the toughest things that you'll see 63 00:04:03,770 --> 00:04:09,260 in machine learning problems but again once we get familiar with this diagram the psychic machine learning 64 00:04:09,260 --> 00:04:16,050 map you'll start to see ah it's actually not as hard as you once might have thought and then again there's 65 00:04:16,050 --> 00:04:20,640 some code that you might use for choosing a model in instantiating one. 66 00:04:20,760 --> 00:04:24,930 Then there's some code here you'll see the functions that we use to fit a model to training data to 67 00:04:24,930 --> 00:04:26,190 make predictions. 68 00:04:26,280 --> 00:04:31,680 We can view our predictions slash predicted probabilities predict the probabilities is done with predict 69 00:04:31,680 --> 00:04:32,370 probe. 70 00:04:33,040 --> 00:04:36,550 And if we come back down here we'll see how we can evaluate our model. 71 00:04:36,780 --> 00:04:41,640 And there's a link here there's a you'll see in the documentation there's a few different ways to evaluate 72 00:04:41,640 --> 00:04:43,470 different socket loan models. 73 00:04:43,470 --> 00:04:48,360 This is a great section I highly suggest reading through here because it's one thing to train a machine 74 00:04:48,360 --> 00:04:48,970 learning model. 75 00:04:48,990 --> 00:04:53,630 It's another thing to check if it's doing the right thing what you want it to do. 76 00:04:53,760 --> 00:04:59,580 Once we've gone through evaluating a model we'll see how we can improve our models through experimentation. 77 00:04:59,580 --> 00:05:04,740 What we're going to focus on is hyper parameter tuning a.k.a. turning the settings and our model and 78 00:05:04,860 --> 00:05:05,680 improving it. 79 00:05:05,760 --> 00:05:11,460 Just like if we wanted to adjust our oven and while oven was cooking our favorite meal maybe it doesn't 80 00:05:11,460 --> 00:05:16,080 do so well on 180 degrees but it does really well on 200 degrees. 81 00:05:16,080 --> 00:05:19,800 The same thing goes with machine learning models out of the box. 82 00:05:19,860 --> 00:05:25,510 It might work pretty well but if you adjust the settings are slightly you might get a better model. 83 00:05:25,560 --> 00:05:27,210 So we'll see that in Section 5. 84 00:05:28,400 --> 00:05:33,560 Then in section six we'll see how to save and reload our trained models so we can avoid going through 85 00:05:33,560 --> 00:05:37,680 all the previous steps before if we wanted to save and share our model. 86 00:05:37,820 --> 00:05:43,340 And then in Step 7 which is kind of a bonus which is not pictured in the workflow is we're gonna put 87 00:05:43,400 --> 00:05:49,790 all of the above steps together in a single cell so that looks like a lot of code at the moment but 88 00:05:49,790 --> 00:05:53,630 don't worry we're gonna step through it by the time you're at the end of this section. 89 00:05:53,660 --> 00:05:55,130 None of this will be unfamiliar. 90 00:05:56,060 --> 00:06:01,130 So with that being said Get yourself ready and get yourself excited pour yourself a cup of coffee or 91 00:06:01,130 --> 00:06:05,300 go take a little walk or something get yourself in the mindset to learn machine learning because that's 92 00:06:05,300 --> 00:06:10,260 what we're going to do we're going to get into the psychic loan library and all of its magical goodness. 93 00:06:10,340 --> 00:06:12,790 So I'll see you in the next video.