WEBVTT 1 00:00:00.350 --> 00:00:01.780 Hi guys and welcome back. 2 00:00:01.780 --> 00:00:02.910 In this video, we're going to learn 3 00:00:02.910 --> 00:00:06.450 how to create a custom legend for our graphs. 4 00:00:06.450 --> 00:00:08.700 We've learned two ways of creating legends. 5 00:00:08.700 --> 00:00:10.550 The first one is to pass in a label 6 00:00:10.550 --> 00:00:14.170 to each plot and then calling .legend on the axes 7 00:00:14.170 --> 00:00:15.950 and that basically generates a legend 8 00:00:15.950 --> 00:00:17.860 for everything that has a label. 9 00:00:17.860 --> 00:00:20.320 The second approach was to assign the result 10 00:00:20.320 --> 00:00:22.160 of calling the plotting functions 11 00:00:22.160 --> 00:00:25.930 to variables and then passing them alongside with labels 12 00:00:25.930 --> 00:00:27.850 to the legends call. 13 00:00:27.850 --> 00:00:31.200 However, what we've got here is a third scenario 14 00:00:31.200 --> 00:00:33.600 where neither of those approaches really work. 15 00:00:33.600 --> 00:00:34.513 So let me explain. 16 00:00:35.350 --> 00:00:38.720 What we've got here is a figure with a single subplot 17 00:00:38.720 --> 00:00:40.060 and then a bar chart. 18 00:00:40.060 --> 00:00:42.180 The bar chart has six columns, 19 00:00:42.180 --> 00:00:43.930 just some random numbers in here 20 00:00:43.930 --> 00:00:47.100 but essentially, the labels are either tech companies 21 00:00:47.100 --> 00:00:48.200 or clothing companies. 22 00:00:48.200 --> 00:00:49.730 So we've got Apple, Google and Microsoft 23 00:00:49.730 --> 00:00:50.660 as tech companies, 24 00:00:50.660 --> 00:00:53.840 Burberry, Zara and Superdry as clothing companies. 25 00:00:53.840 --> 00:00:56.390 Something that we can pass to the bar call 26 00:00:56.390 --> 00:00:57.480 that we haven't looked at yet 27 00:00:57.480 --> 00:00:59.650 is that we can pass a colour list. 28 00:00:59.650 --> 00:01:02.200 Each colour will map to one bar. 29 00:01:02.200 --> 00:01:04.560 So Apple, Google and Microsoft 30 00:01:04.560 --> 00:01:07.390 are gonna be this colour, 5c44fd, 31 00:01:07.390 --> 00:01:08.830 Burberry, Zara and Superdry 32 00:01:08.830 --> 00:01:10.180 are gonna be this colour. 33 00:01:10.180 --> 00:01:11.830 Then we are rotating the xticks 34 00:01:11.830 --> 00:01:12.840 and we're showing this 35 00:01:12.840 --> 00:01:14.750 so let me run it and show you what it looks like. 36 00:01:14.750 --> 00:01:16.000 This is what it looks like. 37 00:01:16.000 --> 00:01:19.030 We've got the blue companies being the tech companies, 38 00:01:19.030 --> 00:01:21.230 red companies being clothing companies. 39 00:01:21.230 --> 00:01:22.890 So we want a legend now 40 00:01:22.890 --> 00:01:26.600 that says blue is tech, red is clothing 41 00:01:26.600 --> 00:01:27.890 but how can we do that 42 00:01:27.890 --> 00:01:29.860 because we only have one plot here? 43 00:01:29.860 --> 00:01:31.200 So if we set a label here, 44 00:01:31.200 --> 00:01:34.030 that's gonna be well, just for the entire plot, 45 00:01:34.030 --> 00:01:36.450 not for the different elements within the plot. 46 00:01:36.450 --> 00:01:38.530 Similarly, if we assign this to a variable 47 00:01:38.530 --> 00:01:40.690 and then we pass it to the legend call, 48 00:01:40.690 --> 00:01:44.280 we again are only gonna be able to set one legend item 49 00:01:44.280 --> 00:01:47.590 for this plot, so we need a custom legend. 50 00:01:47.590 --> 00:01:49.970 Each legend element is composed of two things, 51 00:01:49.970 --> 00:01:52.540 a patch of colour and a label. 52 00:01:52.540 --> 00:01:55.150 So we're going to first of all, create the patch. 53 00:01:55.150 --> 00:01:59.150 So we'll say from matplotlib.patches import Patch. 54 00:01:59.150 --> 00:02:01.670 This allows us to create just a patch of colour somewhere 55 00:02:01.670 --> 00:02:04.900 to be drawn and we're going to add the handles 56 00:02:04.900 --> 00:02:06.640 for the legends down here. 57 00:02:06.640 --> 00:02:08.190 So we'll create two patches. 58 00:02:08.190 --> 00:02:13.190 The first patch has a facecolor of 5c44fd. 59 00:02:13.250 --> 00:02:15.650 And a label of Tech. 60 00:02:15.650 --> 00:02:20.310 And the second patch has a facecolor of ff5566 61 00:02:20.310 --> 00:02:22.270 and a label of Clothing. 62 00:02:22.270 --> 00:02:26.730 So these here are our two handles for the legend. 63 00:02:26.730 --> 00:02:29.500 Handles is the same as a legend item. 64 00:02:29.500 --> 00:02:30.710 So now that we've got this, 65 00:02:30.710 --> 00:02:34.760 all we have to do is say plt.legend handles equal 66 00:02:34.760 --> 00:02:36.950 and we can pass in our variable there. 67 00:02:36.950 --> 00:02:38.280 Now, if we run this, 68 00:02:38.280 --> 00:02:40.240 you can see that we get Tech and Clothing there 69 00:02:40.240 --> 00:02:43.340 with their own little patch of colour and their label. 70 00:02:43.340 --> 00:02:44.350 And this is one of these things 71 00:02:44.350 --> 00:02:45.950 that I mentioned earlier on in the course. 72 00:02:45.950 --> 00:02:48.600 You can see that matplotlib has moved the legend over 73 00:02:48.600 --> 00:02:51.150 to the left because it realised that it would overlap 74 00:02:51.150 --> 00:02:52.670 with this bar here 75 00:02:52.670 --> 00:02:54.790 but it doesn't overlap with anything here. 76 00:02:54.790 --> 00:02:56.450 So that's it, creating a custom legend 77 00:02:56.450 --> 00:02:57.520 can be pretty helpful, 78 00:02:57.520 --> 00:02:59.560 so I hope you've enjoyed this video and learned something. 79 00:02:59.560 --> 00:03:00.570 Thank you for joining me 80 00:03:00.570 --> 00:03:02.220 and I'll see you in the next one.