1 00:00:01,210 --> 00:00:04,920 Let's kick off our exploratory data analysis. 2 00:00:07,390 --> 00:00:09,280 Exploratory data analysis. 3 00:00:09,630 --> 00:00:10,880 And we are going to do two things. 4 00:00:11,190 --> 00:00:13,560 So the first one is that we're going to 5 00:00:13,570 --> 00:00:18,500 do a correlation between caret and price. 6 00:00:19,210 --> 00:00:21,980 And that will be super simple. 7 00:00:22,110 --> 00:00:32,640 So data frame, we get the caret and then we get the price and we do dot core and 8 00:00:32,650 --> 00:00:33,520 here you go. 9 00:00:34,350 --> 00:00:37,460 We see that the correlation is 0 .94 10 00:00:37,470 --> 00:00:39,320 which is huge, right? 11 00:00:40,150 --> 00:00:42,440 Usually there can be an issue when you 12 00:00:42,450 --> 00:00:47,740 have such high correlations because you cannot really distinguish between both 13 00:00:47,750 --> 00:00:53,340 because there's something called mutual causality bias which is when A leads to B 14 00:00:53,350 --> 00:00:54,340 and B leads to A. 15 00:00:55,170 --> 00:00:57,460 However, in this case I think this is 16 00:00:57,470 --> 00:00:59,040 just one direction. 17 00:00:59,270 --> 00:01:02,380 I think the caret leads to the price and 18 00:01:02,390 --> 00:01:07,100 the price does not really lead to the caret because it doesn't make sense that 19 00:01:07,110 --> 00:01:12,220 the bigger the price then the caret increases because you cannot really make 20 00:01:12,230 --> 00:01:13,300 it happen like that. 21 00:01:13,570 --> 00:01:16,060 But if you have a bigger caret then you 22 00:01:16,070 --> 00:01:17,360 have a higher price. 23 00:01:17,950 --> 00:01:21,600 Or in other ways of putting it, you need 24 00:01:21,610 --> 00:01:23,560 to start somewhere, right? 25 00:01:23,630 --> 00:01:25,660 And I think you need to start with having 26 00:01:25,670 --> 00:01:32,260 a bigger caret that will then lead to a higher price because the stone happened 27 00:01:32,270 --> 00:01:33,540 first, not the price. 28 00:01:34,650 --> 00:01:37,160 Therefore, the conclusion is that our 29 00:01:37,170 --> 00:01:43,040 variables are very much tied together which is good from predictability perspective. 30 00:01:43,690 --> 00:01:50,020 Let me isolate x and y because this will be easier as we go through the practice videos. 31 00:01:50,270 --> 00:01:58,800 Isolate x and y. So the y equals our data frame dot price and then the x equals 32 00:01:58,810 --> 00:02:00,960 data frame dot caret. 33 00:02:01,690 --> 00:02:02,420 Here we go. 34 00:02:02,770 --> 00:02:03,400 Control enter. 35 00:02:04,090 --> 00:02:06,240 And I'm getting something. 36 00:02:06,510 --> 00:02:08,980 Object has no attribute price. 37 00:02:09,770 --> 00:02:13,220 That's because it is with a lowercase p 38 00:02:13,230 --> 00:02:14,720 and a lowercase c. 39 00:02:15,830 --> 00:02:18,680 Control enter and here we go. 40 00:02:19,250 --> 00:02:21,620 Now we can do our scatter plot. 41 00:02:22,790 --> 00:02:28,740 In its simplest form scatter plot plt dot 42 00:02:28,750 --> 00:02:36,600 scatter and then we include x and y and then plt dot show. 43 00:02:36,910 --> 00:02:43,100 So this is the simplest form and I can add as well that we are going to 44 00:02:43,110 --> 00:02:47,060 definitely do some or add something to it. 45 00:02:47,430 --> 00:02:51,040 Okay, so it is a lowercase y that I need 46 00:02:51,050 --> 00:02:51,780 to include here. 47 00:02:52,490 --> 00:02:55,780 So this is the simplest form of the 48 00:02:55,790 --> 00:02:57,000 scatter plot. 49 00:02:57,230 --> 00:03:00,820 But let's add a bit to it, right? 50 00:03:00,910 --> 00:03:12,060 So we can add plt dot x label and then this is the caret plt dot y label and 51 00:03:12,070 --> 00:03:20,640 this is the price and then we can add a title plt dot title and then this would 52 00:03:20,650 --> 00:03:27,060 be the relationship of caret and price. 53 00:03:28,770 --> 00:03:29,520 And here we go. 54 00:03:29,910 --> 00:03:32,220 I have something here. 55 00:03:32,470 --> 00:03:36,240 It should be a dot and here we go. 56 00:03:36,290 --> 00:03:40,420 So here I'm customizing the graph. 57 00:03:41,250 --> 00:03:42,500 Here we go. 58 00:03:43,410 --> 00:03:46,440 Control enter and here we go. 59 00:03:46,550 --> 00:03:49,640 So you can see that it's pretty linear, 60 00:03:49,970 --> 00:03:52,740 the relationship, so it really goes in one direction. 61 00:03:53,190 --> 00:03:57,860 You can see that when there is, or when caret equals one, there are a few 62 00:03:57,870 --> 00:04:03,980 outliers, or the more carets you have, it kind of explodes, which would possibly 63 00:04:03,990 --> 00:04:09,340 mean that we actually need a few of the other variables like color clarity or 64 00:04:09,350 --> 00:04:15,800 certification to help us explain these more extreme values where it is exploding. 65 00:04:16,870 --> 00:04:23,760 But one conclusion is that it is pretty linear and this is good to see because it 66 00:04:23,770 --> 00:04:28,980 helps us again to say that we can use a linear regression to predict and it also 67 00:04:28,990 --> 00:04:29,700 makes sense, right? 68 00:04:29,790 --> 00:04:32,200 So the bigger the caret, then caret is 69 00:04:32,210 --> 00:04:36,300 the size of the diamond, then the bigger the price. 70 00:04:36,750 --> 00:04:38,360 So this is also connected. 71 00:04:38,930 --> 00:04:42,100 We then see an inverse relationship somehow. 72 00:04:43,310 --> 00:04:44,800 And this is it. 73 00:04:45,090 --> 00:04:47,720 This is a very simple exploratory data 74 00:04:47,730 --> 00:04:53,000 analysis and in the next video we're going to focus on the linear regression. 75 00:04:53,490 --> 00:04:54,860 Until then, have fun.