1 00:00:00,630 --> 00:00:05,160 I am previous session, we already discussed changing different train parameters. 2 00:00:05,550 --> 00:00:11,220 Now let's check this part to recalculate training, validation and test performance. 3 00:00:12,040 --> 00:00:19,040 If I remember in advance of script number one, I just wrote in a come in window, which was separating 4 00:00:19,170 --> 00:00:25,920 all the indexes that has been used for training the network and also all the indexes and samples which 5 00:00:25,920 --> 00:00:27,990 we use for training and validation. 6 00:00:28,470 --> 00:00:31,310 And we even calculated the error of them. 7 00:00:32,040 --> 00:00:33,450 Let's do the same here. 8 00:00:33,780 --> 00:00:40,350 This time I'm going to create a Wii both for each training, validation and testing so we can call them 9 00:00:40,350 --> 00:00:41,550 whenever we need to. 10 00:00:42,270 --> 00:00:46,740 For the start, let's just separate these three parts. 11 00:00:47,250 --> 00:00:51,060 I will show it as I come in like training plant. 12 00:00:51,990 --> 00:01:05,280 Next, we need validation part and we need another session for testing control X and pay citya control 13 00:01:05,280 --> 00:01:08,610 V for training for validation. 14 00:01:08,820 --> 00:01:09,600 Do the same. 15 00:01:13,980 --> 00:01:23,770 Let's create another variable and name to train index in a train index, I want to put all the training 16 00:01:23,790 --> 00:01:25,280 samples that has been used. 17 00:01:25,380 --> 00:01:32,550 I'm going to use, find function, find every team that is not ñan. 18 00:01:33,180 --> 00:01:40,260 They're in train MSK t are dot train mask. 19 00:01:43,360 --> 00:01:54,370 Don't forget to put this let me go now, do the same for validation data and testing data, but don't 20 00:01:54,370 --> 00:01:57,030 forget to call it out for validation. 21 00:01:57,580 --> 00:02:06,730 And of course, this one is from ask here we have testing decks, a variable that we created, name 22 00:02:06,850 --> 00:02:07,650 test index. 23 00:02:07,660 --> 00:02:09,920 You can just give it any other name that you want. 24 00:02:10,270 --> 00:02:11,140 Find what? 25 00:02:11,140 --> 00:02:14,560 Find anything that is not meant from T. 26 00:02:14,560 --> 00:02:18,010 Are it tested very well. 27 00:02:18,340 --> 00:02:24,750 Now you can already run your program by clicking on the run or you can just type the name of your screen. 28 00:02:25,060 --> 00:02:27,460 Here is Advance is Craib number two. 29 00:02:28,060 --> 00:02:30,100 So I'm going to just click on it here. 30 00:02:30,100 --> 00:02:32,560 We can see that the program run. 31 00:02:33,010 --> 00:02:34,090 I'm going to close it. 32 00:02:34,090 --> 00:02:40,180 And this time let's call for the variable that we created, which is try an index. 33 00:02:40,190 --> 00:02:40,930 Oh, OK. 34 00:02:41,200 --> 00:02:42,970 I just had mistype that. 35 00:02:42,970 --> 00:02:50,800 This is in the the good thing about the come in is, is all of this suggesting a correct word for you 36 00:02:50,800 --> 00:02:51,220 to call? 37 00:02:51,430 --> 00:03:01,240 So the variable that I create called trade I n the X, so here is Train I, D, I and X, just click 38 00:03:01,240 --> 00:03:01,770 on Enter. 39 00:03:02,140 --> 00:03:08,650 These are all the samples that has been used for our training and then the same. 40 00:03:08,650 --> 00:03:10,290 You can just call for that. 41 00:03:10,540 --> 00:03:12,310 Let's see them here. 42 00:03:12,310 --> 00:03:18,370 We have only six samples and as well as for validation index. 43 00:03:20,020 --> 00:03:21,150 OK, here it is. 44 00:03:21,790 --> 00:03:27,370 Now let's back here and change this train together a little bit. 45 00:03:27,790 --> 00:03:30,460 I really don't like this multiplication. 46 00:03:30,460 --> 00:03:37,960 Let me change it to all the rows and from the columns. 47 00:03:37,960 --> 00:03:45,550 Choose that Troian index or the index says that has been used for training in apprentice's, close, 48 00:03:45,790 --> 00:03:48,910 open and close parentheses type here. 49 00:03:49,660 --> 00:03:58,990 All the arrows to mention all the arrows, we can just put a double column and then next parameter. 50 00:03:58,990 --> 00:04:02,770 Here is other columns for the columns. 51 00:04:02,770 --> 00:04:03,700 Which columns? 52 00:04:03,700 --> 00:04:07,060 The train index columns. 53 00:04:07,570 --> 00:04:14,350 This variable that I created here now Velda, to run the program to make sure that we did everything 54 00:04:14,350 --> 00:04:19,630 correct and call for train targets. 55 00:04:21,190 --> 00:04:24,310 Here we can see the list of train targets. 56 00:04:24,310 --> 00:04:25,060 That's better. 57 00:04:25,060 --> 00:04:29,500 Now, let's do the same for validation. 58 00:04:29,500 --> 00:04:37,360 Target open and close parentheses, put a double column and then here instead of train. 59 00:04:37,360 --> 00:04:40,030 This one is of our index. 60 00:04:41,260 --> 00:04:45,970 Please do the same for all test index. 61 00:04:46,670 --> 00:04:47,830 OK, here we go. 62 00:04:48,070 --> 00:04:54,370 We can do the same for inputs, for output, the output of our neural network. 63 00:04:54,550 --> 00:04:56,590 Let's see how would it happen here. 64 00:04:56,980 --> 00:05:06,070 Just copy this part pasted again for inputs and pasted it one more time for this time outputs and let 65 00:05:06,070 --> 00:05:07,000 me change the name. 66 00:05:07,000 --> 00:05:08,980 This is train inputs. 67 00:05:10,930 --> 00:05:11,950 Train inputs. 68 00:05:12,100 --> 00:05:23,590 This one is train out puts my inputs in this program called X and my outputs are called Y. 69 00:05:23,590 --> 00:05:26,280 Y is the output of neural network. 70 00:05:26,770 --> 00:05:31,060 Please do the same for your validation and testing. 71 00:05:31,900 --> 00:05:33,070 OK, here we go. 72 00:05:33,070 --> 00:05:36,100 I already changed them for validation and testing. 73 00:05:36,100 --> 00:05:44,170 I just added test inputs, test outputs as well as validation inputs and validation outputs. 74 00:05:44,590 --> 00:05:48,370 Just copy, paste them and change them into X and Y. 75 00:05:49,720 --> 00:05:51,160 What else can we do here? 76 00:05:51,310 --> 00:05:53,260 Actually, there are lots of things to do. 77 00:05:53,560 --> 00:06:00,760 For example, we can even calculate the train errors for the training errors. 78 00:06:00,760 --> 00:06:05,290 I'm going to define another variable I'm calling train errors. 79 00:06:06,310 --> 00:06:09,160 This one is equal to train targets. 80 00:06:10,660 --> 00:06:20,470 This variable minus the train outputs the output of our neural network variable. 81 00:06:21,700 --> 00:06:25,570 You can calculate the error for each of them. 82 00:06:25,570 --> 00:06:29,770 You can just do it for the validation and you can do it for the testing. 83 00:06:31,670 --> 00:06:38,300 So there we have the validation errors, we have the test errors, and this one is validation targets 84 00:06:38,300 --> 00:06:42,230 minus the validation out, which is the same as testing. 85 00:06:42,800 --> 00:06:47,060 Now, let me run the program to see if we did everything correct. 86 00:06:47,630 --> 00:06:51,010 OK, it turns out we have a problem here. 87 00:06:51,260 --> 00:06:58,310 Line 81, let's check the problem together now is undefined function or variable. 88 00:06:58,670 --> 00:07:05,870 I just missed time that this one is out p you t is now I everything correct. 89 00:07:06,320 --> 00:07:06,890 Let's see. 90 00:07:06,890 --> 00:07:08,090 Yeah, here we have. 91 00:07:08,990 --> 00:07:18,290 OK, now let's call for the test errors to see one or two errors for the test here we have the errors 92 00:07:18,500 --> 00:07:21,950 but we have train errors. 93 00:07:22,460 --> 00:07:26,840 We can't see them here and we can have the validation errors. 94 00:07:27,290 --> 00:07:31,550 So you can just separate your data for better analysis. 95 00:07:31,700 --> 00:07:34,720 We have over outputs, we have the inputs. 96 00:07:34,730 --> 00:07:39,020 Let me just show the train inputs. 97 00:07:39,380 --> 00:07:46,250 These are the samples that we used for input to train our network and then you can just call for any 98 00:07:46,250 --> 00:07:46,660 of them. 99 00:07:47,600 --> 00:07:54,680 So we already covered them, recalculating training, validation and test performance for the test performance. 100 00:07:54,680 --> 00:07:55,820 You have your codes here. 101 00:07:56,030 --> 00:08:02,510 If you need them, just put a semicolon at the end of each one and then uncommented them. 102 00:08:03,260 --> 00:08:05,630 Next part would be plotts. 103 00:08:05,870 --> 00:08:09,950 I'm going to just give a similar explanation for diplomats in the next video. 104 00:08:10,520 --> 00:08:11,240 So you there?