WEBVTT 00:00.000 --> 00:07.400 Hey folks, welcome to notebook 2.0 where we'll start really diving into machine learning. 00:07.400 --> 00:14.160 And I hope that the notebook 1 and 1.5 were a good preparation, good warm up, good refresher 00:14.160 --> 00:17.800 on your data science skills. 00:17.800 --> 00:21.000 Before we actually get started, I do want to make sure we're all set up with our notebook. 00:21.000 --> 00:26.560 We don't want to be wondering whether what's going wrong is our notebook infrastructure or 00:26.560 --> 00:27.880 our code. 00:27.880 --> 00:33.040 So I'm going to run this, it's already mounted, but for you, it's probably going to come 00:33.040 --> 00:40.760 with a whole set of dialogue boxes for configuring Google Drive, connecting it to this particular 00:40.760 --> 00:41.760 notebook. 00:41.760 --> 00:44.000 It also will take a moment to just start running. 00:44.000 --> 00:47.200 Let's also run our import blocks. 00:47.200 --> 00:51.360 Here's this one and this one. 00:51.360 --> 00:57.200 And we're importing a lot of our typical NumPy, pandas, and pi plot. 00:57.200 --> 01:02.800 Also Seaborn, we'll be using that and our ability to display images. 01:02.800 --> 01:07.280 And for the first time, we're going to import some things from Scikit Learn, which is a 01:07.280 --> 01:11.160 science library devoted to machine learning. 01:11.160 --> 01:15.520 And we're going to be importing a linear regression model, which is different from our linear 01:15.520 --> 01:18.400 regression formulas. 01:18.400 --> 01:22.520 The ability to split data into testing and training will be talking about that in a moment. 01:22.520 --> 01:28.080 And Standard Scaler is a tool Scikit Learn has for standardizing our data. 01:28.080 --> 01:33.520 All right, so make sure you've run all three of those so that you're all set up. 01:33.520 --> 01:39.560 And now we'll go and talk about what we're doing here. 01:39.560 --> 01:44.560 So supervised machine learning, that's what we're doing here. 01:44.560 --> 01:48.520 It involves helping an algorithm learn. 01:48.520 --> 01:50.320 And we'll talk a lot about what that means, exactly. 01:50.320 --> 01:52.480 What does it mean for a model to learn? 01:52.480 --> 01:57.320 And we're going to train it based on some data. 01:57.320 --> 02:00.120 So we'll train it to make a prediction. 02:00.120 --> 02:03.640 It'll provide an output, which is called a prediction. 02:03.640 --> 02:09.640 And this phrase gets people a little bit. 02:09.640 --> 02:12.440 We're going to be talking about features, different features. 02:12.440 --> 02:14.320 And those are just inputs. 02:14.320 --> 02:20.480 So think columns in a data frame, like which of these characteristics of our data are 02:20.480 --> 02:21.480 important. 02:21.480 --> 02:25.280 Those are the ones we'll feed to the model as features. 02:25.280 --> 02:30.960 Now when the model makes a prediction, when we train the model, we'll give it the features, 02:30.960 --> 02:34.440 we'll give it all of the data that we wanted to train on. 02:34.440 --> 02:37.920 And we'll also give it the answers, the target. 02:37.920 --> 02:45.000 So the first thing we'll be doing, for example, is trying to predict car sales numbers. 02:45.000 --> 02:46.920 How much will this car sell for? 02:46.920 --> 02:49.440 What will it price be? 02:49.440 --> 02:56.040 Given things like horsepower, engine size, fuel efficiency, and several other elements 02:56.040 --> 02:58.160 of a car. 02:58.160 --> 03:05.600 Now what will give it is, hey, here is all the data we think could be helpful for predicting 03:05.600 --> 03:08.240 our car's data. 03:08.240 --> 03:10.720 And also, here's the prices. 03:10.720 --> 03:11.720 Here's the answers. 03:11.720 --> 03:17.080 So we can learn from that, hey, given x and y, z was the answer. 03:17.160 --> 03:20.960 And this x and y, this z was the answer. 03:20.960 --> 03:26.040 So the correct purpose of the goal is for it to be able to predict it, and it needs 03:26.040 --> 03:31.080 to know, hey, what are the general values that I can predict it based on, and I can reverse 03:31.080 --> 03:40.680 engineer exactly what model, exactly what formula will get me from the inputs to the output. 03:40.680 --> 03:43.440 And we give also called the correct answer, the label. 03:43.440 --> 03:46.120 Since we've labeled the data, this is the answer. 03:46.680 --> 03:49.560 This is this answer, et cetera. 03:49.560 --> 03:52.800 Label will make even more sense when it gets a classifying data. 03:52.800 --> 03:55.760 But we'll see that in a little while. 03:55.760 --> 03:58.600 But it'll make more sense as a term for that. 03:58.600 --> 04:05.160 OK, so we're going to give it training data, inputs, paired with the correct label with 04:05.160 --> 04:09.000 features, and it will give it the correct answer. 04:09.000 --> 04:13.560 Now when we train it, we'll give it these characteristics, fuel efficiency, engine size 04:13.560 --> 04:20.760 horsepower, and typically it's like one row per car, right, one set of data. 04:20.760 --> 04:26.720 And also the label, the correct answer, hey, what was the price for that car? 04:26.720 --> 04:35.720 And then we hope that it will be able to learn, hey, what features, what inputs, what values 04:35.720 --> 04:42.400 those features lead to, what values of prices of our target. 04:42.400 --> 04:50.320 So that's for cars, if we wanted a classification, we would have dogs and cats, pictures of dogs 04:50.320 --> 04:51.320 and cats. 04:51.320 --> 04:52.720 And the features would be those images. 04:52.720 --> 04:55.240 And the labels, which again makes more sense now that we're classifying. 04:55.240 --> 04:58.040 We labeled this one a dog, this one a cat. 04:58.040 --> 05:02.360 And if the model sees enough pictures of dogs and cats, it'll learn, hey, what about 05:02.360 --> 05:04.520 this picture makes it a cat? 05:04.520 --> 05:08.240 What about this picture makes it a dog? 05:08.240 --> 05:11.960 Now it's not seeing season quotes here. 05:11.960 --> 05:13.400 It's just looking at numbers. 05:13.400 --> 05:18.160 It's just looking at a Python list of numbers, typically. 05:18.160 --> 05:25.240 So a colorful photo of a cat would actually be a whole list of pixels, red, green, and 05:25.240 --> 05:26.240 blue values. 05:26.240 --> 05:28.000 How much red, how much green, how much blue? 05:28.000 --> 05:30.040 Because it's just numbers to the machine. 05:30.040 --> 05:35.040 It doesn't know cats, it doesn't know dogs, it just knows data. 05:35.040 --> 05:37.840 And all outputs are going to be made, are going to be numbers too. 05:37.840 --> 05:42.600 So cat versus dog, we might label cat the number one dog, the number two. 05:42.600 --> 05:49.920 And it says, hey, when I see this set of pixels that look a little bit like this, that looks 05:49.920 --> 05:54.800 like it goes with one, when I see a pixels that have this, these numbers in them, that 05:54.800 --> 05:57.360 looks like it goes for with two. 05:57.360 --> 06:01.840 So then I can predict, oh, this one looks like a one, this one looks like a two. 06:01.840 --> 06:05.680 We're also going to split off our data, not just in a training data, but into testing 06:05.680 --> 06:07.160 data as well. 06:07.160 --> 06:11.760 This data will be, will withhold from the model, we won't give it all. 06:11.760 --> 06:18.000 Instead we will hold data in reserve so that we can say, okay, you learned on this data, 06:18.000 --> 06:22.240 now here's a quiz, here's the test data. 06:22.240 --> 06:28.240 How good are you given this, given what you learned, run your predictions on the features 06:28.240 --> 06:35.120 of this test data, on the inputs, on the values we give you, and what's your prediction? 06:35.120 --> 06:39.640 And we'll measure that against the actual results for that testing data. 06:39.640 --> 06:42.760 So you know, this is like, okay, when you saw a dog, you want to solve it, a dog saw what 06:42.760 --> 06:47.240 a cat you think you got the idea, here's a picture of a dog, here's a picture of a cat, 06:47.240 --> 06:52.360 which one do you think is this a cat or a dog, is this a cat or a dog, and it's never seen 06:52.360 --> 06:54.840 those exact cats or dogs before. 06:54.840 --> 06:58.160 It's never seen these cars before, it doesn't know what prices they are. 06:58.160 --> 07:03.160 And if it's learned, it's less and well, it'll do a good job of predicting it, predicting 07:03.160 --> 07:05.320 new data based on the old one. 07:05.320 --> 07:08.240 So it wouldn't be very helpful if it was just like, yeah, I've seen that picture, you told 07:08.240 --> 07:10.680 me it was a cat, I remember it, that's a cat. 07:10.680 --> 07:14.360 You showed me this picture of a dog, I remember it's a dog, I remember you told me, it's 07:14.360 --> 07:17.200 like, okay, we'll show you a new picture, it's like, I've never seen that before, but 07:17.200 --> 07:21.920 I think based on what I've seen a cat's a dog, that one's a dog, and you're like, oh, 07:21.920 --> 07:25.240 no, that one's a cat, oh, sorry, heartbreaker. 07:25.240 --> 07:28.720 But hopefully our data, our model will be pretty good at it. 07:28.720 --> 07:32.520 And generally, we're going to split it into a data set testing and training. 07:32.520 --> 07:38.440 And typically, 80% of the data we use training and 20% testing. 07:38.440 --> 07:41.840 That's what people have found is very helpful. 07:41.840 --> 07:44.480 All right, that's our introduction. 07:44.480 --> 07:45.000 Let's get started.