1 00:00:00,800 --> 00:00:07,400 Highlight continue with our project, and it is we will do some neural network model sortable for my 2 00:00:07,400 --> 00:00:09,120 neural network regression analysis. 3 00:00:09,170 --> 00:00:11,390 We will do it across sequential model. 4 00:00:11,990 --> 00:00:20,150 So there are some force that you need to do to perform the model so that some Stanford also. 5 00:00:21,170 --> 00:00:24,650 Certainly tomorrow, and the first that a. 6 00:00:25,830 --> 00:00:26,490 Import. 7 00:00:28,840 --> 00:00:30,310 Sequential. 8 00:00:33,070 --> 00:00:38,260 From Kiraz DOT models, the next one is. 9 00:00:39,640 --> 00:00:44,710 Stock the life using DOT at. 10 00:00:46,430 --> 00:00:47,060 Michael. 11 00:00:49,030 --> 00:00:50,170 The next one is. 12 00:00:51,240 --> 00:00:59,790 Configure the learning process using DOT compile. 13 00:01:06,670 --> 00:01:08,830 Method, and the last one is. 14 00:01:12,230 --> 00:01:14,690 Trained model anda. 15 00:01:16,210 --> 00:01:19,660 Trade data said using. 16 00:01:21,310 --> 00:01:22,930 Don't fit. 17 00:01:24,320 --> 00:01:24,920 Madha. 18 00:01:26,920 --> 00:01:33,780 So we have made the import data step before, so we call our data is already available. 19 00:01:34,450 --> 00:01:35,460 We raise you to. 20 00:01:38,400 --> 00:01:41,200 Following steps, are we right? 21 00:01:41,230 --> 00:01:42,060 I'm Cofferati. 22 00:01:42,120 --> 00:01:47,520 So first we have to import from Usdaw model. 23 00:01:49,980 --> 00:01:50,670 In part. 24 00:01:55,740 --> 00:01:56,580 Cranshaw. 25 00:01:59,140 --> 00:02:04,510 And from cross dot dildoes import. 26 00:02:06,220 --> 00:02:06,880 Dence. 27 00:02:08,960 --> 00:02:13,920 And then from Kiraz, not magic. 28 00:02:16,420 --> 00:02:21,070 From Keri's import matrix. 29 00:02:23,350 --> 00:02:34,000 So to be happy about that which are sequential and metric so that I am not what I recall, they are 30 00:02:34,000 --> 00:02:34,900 very important. 31 00:02:39,770 --> 00:02:40,460 So. 32 00:02:43,820 --> 00:02:46,130 Sequential glass. 33 00:02:48,000 --> 00:02:51,840 This a usual Devi. 34 00:02:54,110 --> 00:02:56,540 Linear stock of. 35 00:02:59,170 --> 00:02:59,830 Network. 36 00:03:00,740 --> 00:03:06,110 They are not a model. 37 00:03:10,310 --> 00:03:12,800 We will show you the. 38 00:03:13,820 --> 00:03:15,320 Sequential. 39 00:03:16,710 --> 00:03:20,850 Constructor to create a model. 40 00:03:23,080 --> 00:03:25,210 Which will be. 41 00:03:26,610 --> 00:03:30,400 Inrush would play a. 42 00:03:31,750 --> 00:03:32,590 Using. 43 00:03:33,830 --> 00:03:35,420 They add. 44 00:03:36,420 --> 00:03:37,040 Method. 45 00:03:38,880 --> 00:03:39,870 That as glass. 46 00:03:44,670 --> 00:03:48,650 There's a huge hole in charge here. 47 00:03:49,580 --> 00:03:50,160 Uh. 48 00:03:53,750 --> 00:03:54,480 They are. 49 00:03:55,970 --> 00:03:59,030 Which is the basic. 50 00:04:02,920 --> 00:04:03,670 For. 51 00:04:05,130 --> 00:04:08,990 Fully connected, the. 52 00:04:12,160 --> 00:04:15,010 And then the final is Metrick. 53 00:04:16,900 --> 00:04:18,130 Class, so. 54 00:04:19,300 --> 00:04:22,150 The ADA is a function. 55 00:04:23,880 --> 00:04:33,000 That is due to evaluate the performance of the model. 56 00:04:38,220 --> 00:04:41,340 So now let's start to write something called. 57 00:04:44,300 --> 00:04:47,060 Next one will be Modahl. 58 00:04:49,870 --> 00:04:50,580 Equa. 59 00:04:52,330 --> 00:04:53,470 Sequential. 60 00:04:56,700 --> 00:05:00,860 Modahl, not at this. 61 00:05:05,280 --> 00:05:07,890 20 Khama. 62 00:05:09,170 --> 00:05:11,450 In both underscore the. 63 00:05:13,510 --> 00:05:15,940 And the dimension of the 2013. 64 00:05:17,510 --> 00:05:19,280 And activation. 65 00:05:21,190 --> 00:05:21,910 Equal. 66 00:05:23,130 --> 00:05:23,700 Relu. 67 00:05:25,500 --> 00:05:26,970 Then add one more. 68 00:05:32,060 --> 00:05:32,600 There's. 69 00:05:34,560 --> 00:05:37,680 Ten input equa. 70 00:05:39,820 --> 00:05:44,050 They're all because this is our first hidden layer. 71 00:05:46,290 --> 00:05:59,070 And the last one is this is one this one will be the award show activation with A'Lelia. 72 00:06:08,430 --> 00:06:16,260 So we're going to say that for the first role, we have set a to model a sequential and then we have 73 00:06:16,260 --> 00:06:17,270 added Nalliah. 74 00:06:19,310 --> 00:06:19,970 So are we. 75 00:06:20,030 --> 00:06:20,400 We were. 76 00:06:20,420 --> 00:06:26,150 You were completely correct, that nanostructure, which will us fully connected level advisers, the 77 00:06:26,480 --> 00:06:33,770 glass sort of the most important because it's a laboratory fighter, a boat, it might have the right 78 00:06:33,770 --> 00:06:34,210 number. 79 00:06:34,220 --> 00:06:43,670 It was so because we had 13 rows that far away, we had to set in to be 13, 13 predictors. 80 00:06:44,330 --> 00:06:46,850 And we Balestrieri by Ramatoulaye 20. 81 00:06:49,360 --> 00:06:51,940 In beauty and activation. 82 00:06:54,970 --> 00:07:00,670 So the trend is a positive integer, probably ending the dimensionality of the abuse by. 83 00:07:01,630 --> 00:07:05,410 And they note the number of neurons on the inner level. 84 00:07:07,660 --> 00:07:16,020 In what they attribute is the number in both variables and finally the activation is equal, RELU action 85 00:07:16,410 --> 00:07:18,380 is to set the activation function. 86 00:07:18,870 --> 00:07:23,830 The second layer has 10 neurons and RELU activation function. 87 00:07:24,600 --> 00:07:26,170 So this one is trending around. 88 00:07:26,190 --> 00:07:27,800 This one is ten neurons. 89 00:07:28,590 --> 00:07:30,060 And finally, the outlier. 90 00:07:30,060 --> 00:07:33,630 I had only one neuron and the activation function is linear. 91 00:07:34,870 --> 00:07:35,940 So we don't have. 92 00:07:37,650 --> 00:07:39,310 An all time to talk about a. 93 00:07:40,230 --> 00:07:48,540 Activation version, I did explain very clearly on my auto courses, which is I in healthcare, so. 94 00:07:49,760 --> 00:07:51,560 Now before training. 95 00:07:52,490 --> 00:07:59,780 Before training a model, you need to configure the learning process, which is a combined method so 96 00:07:59,780 --> 00:08:00,890 that at. 97 00:08:02,940 --> 00:08:06,150 They're going to say who say it, we didn't make any mistake. 98 00:08:10,040 --> 00:08:16,620 And I forgot to run this out, that's why and we got no mistake. 99 00:08:17,150 --> 00:08:21,040 So it's very easy to come by. 100 00:08:24,280 --> 00:08:27,680 Up the Mizar equal idam. 101 00:08:31,180 --> 00:08:31,660 Los. 102 00:08:33,890 --> 00:08:34,610 Equa. 103 00:08:35,810 --> 00:08:44,070 I mean, underscores why underscore error and then the last one is. 104 00:08:45,280 --> 00:08:45,880 Metrick. 105 00:08:48,440 --> 00:08:49,190 Equa. 106 00:08:52,360 --> 00:08:53,050 Racey. 107 00:08:56,580 --> 00:09:02,500 So that horrible so that come by for us and then they talk about it. 108 00:09:02,520 --> 00:09:06,640 So the first one is I optimizer. 109 00:09:07,440 --> 00:09:08,070 So. 110 00:09:09,110 --> 00:09:09,720 Adam. 111 00:09:11,170 --> 00:09:13,040 Up the Mizar. 112 00:09:13,270 --> 00:09:15,480 So why is it this age? 113 00:09:16,760 --> 00:09:17,600 An al. 114 00:09:19,240 --> 00:09:24,430 We are for the first order. 115 00:09:27,130 --> 00:09:37,210 Gradient by optimization of stochastic objective. 116 00:09:38,530 --> 00:09:39,250 Function. 117 00:09:40,610 --> 00:09:48,230 Based on duct tape, estimates of lower order. 118 00:09:50,890 --> 00:09:51,550 Moman. 119 00:09:52,710 --> 00:09:57,090 The next one is the maze where Soviet era. 120 00:09:58,190 --> 00:09:59,880 Of course, we want to minimize. 121 00:10:01,990 --> 00:10:03,620 We want our model is very good. 122 00:10:07,560 --> 00:10:08,880 Which is the lowest. 123 00:10:10,280 --> 00:10:11,030 Function. 124 00:10:12,710 --> 00:10:23,390 So it measured the average of the square of the. 125 00:10:25,460 --> 00:10:26,210 Arroz. 126 00:10:27,180 --> 00:10:29,520 So the last one is. 127 00:10:30,530 --> 00:10:31,190 The. 128 00:10:33,110 --> 00:10:40,070 These are Kilrea, see Mattrick, which is a function. 129 00:10:41,860 --> 00:10:45,490 That is used to evaluate. 130 00:10:47,150 --> 00:10:47,720 The. 131 00:10:48,640 --> 00:10:51,010 Well, foreman of the model. 132 00:10:52,250 --> 00:10:52,940 Doray. 133 00:10:55,170 --> 00:10:57,720 The training and testing. 134 00:11:00,620 --> 00:11:02,060 So now we got you. 135 00:11:07,630 --> 00:11:08,560 More out of it. 136 00:11:11,080 --> 00:11:13,390 My thoughts are Modot dog fit. 137 00:11:15,380 --> 00:11:16,270 Then as. 138 00:11:18,300 --> 00:11:21,270 Train, comma, why? 139 00:11:22,480 --> 00:11:22,960 Train. 140 00:11:25,090 --> 00:11:26,020 And Epoch. 141 00:11:27,280 --> 00:11:30,340 A 1000 and proposed. 142 00:11:31,340 --> 00:11:39,920 Which is one, so, yeah, we don't have time to go through all the definition in here, but if you 143 00:11:39,920 --> 00:11:46,370 can, how will we explain or I have explained very carefully on my order courses, which had five big 144 00:11:46,370 --> 00:11:49,030 projects we got than this in health care. 145 00:11:49,730 --> 00:11:53,110 So now that when our. 146 00:11:54,520 --> 00:11:58,330 Neuron and eyes are far, far away because it will take a long time. 147 00:12:00,350 --> 00:12:05,540 I did trained Amoroso assembly, and now we need to bring the model summary to say. 148 00:12:07,520 --> 00:12:08,480 What is happening? 149 00:12:11,700 --> 00:12:17,280 I secured a sale and we got a 501 Paretta. 150 00:12:19,640 --> 00:12:28,550 So here we clearly say the opposite and no surprises in life, so to test the model capacity for predicting 151 00:12:28,550 --> 00:12:35,600 the median value of owner occupied homes as a function of the 13 variables has shown, we can display 152 00:12:35,600 --> 00:12:38,610 the predicted values against the current one. 153 00:12:39,200 --> 00:12:46,340 So to do this, we must perform the prediction on all the observation contending that has by the predict 154 00:12:46,340 --> 00:12:46,850 function. 155 00:12:48,050 --> 00:12:51,770 So it's very easy with that Cranio called. 156 00:12:53,260 --> 00:12:55,780 Tell you why underscore. 157 00:12:56,760 --> 00:12:59,340 That you its corporate. 158 00:13:00,520 --> 00:13:01,150 K.M.. 159 00:13:03,120 --> 00:13:05,820 Equal model, not predict. 160 00:13:07,710 --> 00:13:08,520 And then. 161 00:13:09,390 --> 00:13:13,390 I underscored tests that run the sale. 162 00:13:14,610 --> 00:13:15,150 And. 163 00:13:16,550 --> 00:13:22,430 We did not get any error, so for the time being, we don't need to add anything else, the that will 164 00:13:22,430 --> 00:13:26,510 be useful for us to make a comparison with another model. 165 00:13:26,900 --> 00:13:35,660 So to evaluate our performance, we must use to evaluate function so that Antonio Cazale and score. 166 00:13:37,650 --> 00:13:41,130 Equal model or evaluate. 167 00:13:43,630 --> 00:13:45,130 I underscored his. 168 00:13:46,430 --> 00:13:55,640 Why underscore the test and the polls equals zero and then just bring. 169 00:13:59,150 --> 00:13:59,990 Carus. 170 00:14:01,360 --> 00:14:02,020 Model. 171 00:14:03,370 --> 00:14:04,780 And then bring the skull. 172 00:14:06,110 --> 00:14:08,240 Score zero. 173 00:14:10,500 --> 00:14:15,990 And this one said, A had lost value and magic value for the modern enticement. 174 00:14:17,140 --> 00:14:20,660 So can both computation is done in patches. 175 00:14:21,640 --> 00:14:22,060 So. 176 00:14:23,290 --> 00:14:31,360 That run the sale and we got zero point zero six, so as workmen said, to be a good result, but to 177 00:14:31,360 --> 00:14:39,460 have a confirmation necessary to compare these results with those deriving from the application or another 178 00:14:39,460 --> 00:14:45,250 model, as we are doing regression analysis, this model seemed to be the most suitable one. 179 00:14:45,730 --> 00:14:47,980 And this is the end of this video. 180 00:14:48,250 --> 00:14:49,360 I hope you enjoy it. 181 00:14:49,690 --> 00:14:51,640 And I will see you in the next video.