1 00:00:01,350 --> 00:00:06,430 All right now that you know how to do slicing and iterating through Numpy arrays 2 00:00:06,420 --> 00:00:13,090 let's do some more specific operations and would be stacking Nympy arrays so concatenating 3 00:00:13,090 --> 00:00:19,960 Numpy arrays to each other and also splitting a Numpy array to smaller arrays. 4 00:00:20,360 --> 00:00:23,560 I have this Numpy array still here. 5 00:00:23,640 --> 00:00:29,680 So let's start with stocking to Numpy arrays. 6 00:00:29,700 --> 00:00:36,510 To do that you may want to create a new variable where you will save your big array and that would be equal to Numpy 7 00:00:37,050 --> 00:00:39,850 and the method to stock two Numpy arrays� 8 00:00:40,020 --> 00:00:48,180 Actually there are two methods one with is horizontal stack so hstack. That expects for you to pass 9 00:00:48,300 --> 00:00:52,240 two or more number arrays that you want to stack horizontally. 10 00:00:52,280 --> 00:00:56,370 So let me pass. 11 00:00:56,460 --> 00:00:57,080 Im_g 12 00:00:57,090 --> 00:00:59,880 and then the same array. 13 00:00:59,880 --> 00:01:05,310 It's not a problem. If you have two different arrays you can you could do that but I'm just passing the 14 00:01:05,310 --> 00:01:06,630 same array there. 15 00:01:06,630 --> 00:01:08,710 Now if you execute this you'll get an error. 16 00:01:08,820 --> 00:01:12,540 And I want you to see this error so that you understand it. 17 00:01:12,660 --> 00:01:17,640 So it says hstack takes 1 positional argument but 2 were given. 18 00:01:17,700 --> 00:01:19,730 Which means that the hstack 19 00:01:19,770 --> 00:01:25,260 inside the brackets it gets only 1 argument but we are passing 2. 20 00:01:25,260 --> 00:01:29,430 So how do we go about concatenating 2 Numpy arrays? 21 00:01:29,460 --> 00:01:31,740 We can't just pass a Numpy array there. 22 00:01:32,040 --> 00:01:41,180 So the solution here is to have a tuple of Numpy arrays. Just like that. 23 00:01:41,190 --> 00:01:50,700 So a tuple inside the input or a hstack method. You execute that. 24 00:01:51,180 --> 00:01:54,220 And here is a stacked array. Yes, so 25 00:01:54,240 --> 00:02:00,810 this is the first array and then the next array which in this case happens to be the same one. 26 00:02:00,810 --> 00:02:04,470 But if you want you can add more 27 00:02:09,360 --> 00:02:13,250 and you get a longer array which is not very good looking like that. 28 00:02:13,440 --> 00:02:19,060 So you can't print that out and you know you see the difference. 29 00:02:19,360 --> 00:02:24,300 So with the first array, the second and the third one. 30 00:02:27,030 --> 00:02:32,280 And if you want to stack vertically you'd want to do this there. 31 00:02:34,900 --> 00:02:40,020 And yeah, as you expected 32 00:02:40,020 --> 00:02:47,460 this will concatenate the arrays in the vertical position, in the vertical axis. 33 00:02:47,550 --> 00:02:48,340 Be aware that 34 00:02:48,390 --> 00:02:56,470 if you try to concatenate arrays that have different dimensions you'll have, 35 00:02:56,560 --> 00:02:57,610 you'll get an error. 36 00:02:57,630 --> 00:03:00,650 So Ims has three dimensions 37 00:03:00,660 --> 00:03:09,150 if you can remember that from up here. Now you learnd how to concatenate. And how both splitting an array 38 00:03:09,150 --> 00:03:11,700 into smaller arrays? 39 00:03:12,180 --> 00:03:21,440 Well to do that you could create a variable and then Numpy.horizontalsplit for splitting horizontally 40 00:03:22,310 --> 00:03:25,330 and then this expects a Numpy array. 41 00:03:25,500 --> 00:03:28,860 And how many arrays you want to produce out of that? 42 00:03:29,340 --> 00:03:35,400 So horizontally let's say 3. 43 00:03:37,310 --> 00:03:42,870 Yeah, this says array spilt does not result in an equal division. 44 00:03:42,870 --> 00:03:46,640 The reason for that is you know we're trying� 45 00:03:46,710 --> 00:03:50,170 This array has one two three four five columns. 46 00:03:50,220 --> 00:03:56,030 So you're trying to divide five with three and Numpy cannot decide how to do that. 47 00:03:56,220 --> 00:04:03,300 So should it give to you a Numpy array with this 2 columns first and then another Numpy array with two columns 48 00:04:03,660 --> 00:04:05,920 and then the last Numpy array with one column. 49 00:04:05,970 --> 00:04:08,720 So Numpy cannot do that. 50 00:04:08,810 --> 00:04:12,630 Well you should do is maybe 5 there 51 00:04:13,140 --> 00:04:15,530 and lst. 52 00:04:18,390 --> 00:04:24,930 That should give you five different Numpy arrays which in this case happens to be you know one array for 53 00:04:24,960 --> 00:04:33,930 each column and you can do vertical splitting. This time we can try three because we have nine rows there. 54 00:04:36,870 --> 00:04:41,640 And yeah, we get three small arrays. 55 00:04:42,090 --> 00:04:44,640 Note that this lst, so the split 56 00:04:44,700 --> 00:04:49,090 what it produces is Python list of Numpy arrays. 57 00:04:50,010 --> 00:04:58,210 So it's a playing Python list and out of that you can access each of the Numpy arrays if you like. 58 00:04:58,520 --> 00:05:02,770 You know that. It outputs the first arrey. 59 00:05:03,170 --> 00:05:10,020 And yeah that's about concatenating and splitting Numpy arrays. Hope I've given you a good overview of Numpy. 60 00:05:10,040 --> 00:05:10,320 61 00:05:10,740 --> 00:05:18,390 I try to do some practical examples like opening images but this is still yet in the theoretical part. 62 00:05:18,390 --> 00:05:24,730 Numpy is quite theoretical and you only understand it when you get to use it like these. 63 00:05:24,800 --> 00:05:31,080 Real world examples like working with Pandas as we did previously in the course and also working with 64 00:05:31,080 --> 00:05:32,220 images. 65 00:05:32,220 --> 00:05:35,090 So we've run into some hard core image processing 66 00:05:35,110 --> 00:05:39,190 in the next lectures, in the next session and you'll see how 67 00:05:39,180 --> 00:05:41,590 Numpy comes in handy in there. 68 00:05:41,760 --> 00:05:42,950 And yeah, I hope you enjoyed this. 69 00:05:42,960 --> 00:05:44,220 I'll see you later.