1 00:00:00,150 --> 00:00:07,580 Hello and welcome to this lecture, where we are going to solve the flight schedule problem using the 2 00:00:07,710 --> 00:00:08,430 library. 3 00:00:08,610 --> 00:00:18,430 Let's start by defining the toolbox to initialize the library Bayes dots to box. 4 00:00:18,870 --> 00:00:23,400 Then we need to create the fitness function and also the individual. 5 00:00:23,910 --> 00:00:24,770 Let's go. 6 00:00:25,190 --> 00:00:36,750 Greater blots create a yellowing needs to type fitness meme because we have a minimization problem. 7 00:00:37,050 --> 00:00:41,100 We need to get the lowest possible price. 8 00:00:41,580 --> 00:00:47,550 Then we need to type base, but fitness again. 9 00:00:47,970 --> 00:00:57,120 We have with the weights and as we have on minimization problem, we need to put two here minus one 10 00:00:57,510 --> 00:01:02,850 when we have a maximization problem, which puts positive one. 11 00:01:03,870 --> 00:01:14,880 Now let's create the individual's grade through dots, create the name of the class individual. 12 00:01:15,570 --> 00:01:19,620 It will be in the format of a list the next. 13 00:01:21,660 --> 00:01:32,520 Fitness equals greater dot fitness min again because we have a minimization problem. 14 00:01:33,000 --> 00:01:35,760 Now we need to register some objects. 15 00:01:36,660 --> 00:01:43,890 The box dots register before we had a Boolean attributes. 16 00:01:44,250 --> 00:01:56,280 Now we have three composed of integer values because as we can see here in the solution, it is composed 17 00:01:56,310 --> 00:02:07,410 of numbers in the range from zero to nine because that are then available flights for each one of this 18 00:02:07,410 --> 00:02:08,040 series. 19 00:02:08,670 --> 00:02:12,960 So now we need to define an integrate attribute. 20 00:02:13,410 --> 00:02:23,970 Let's go again random wrens eons to generate the random numbers and that are two other parameters a 21 00:02:24,450 --> 00:02:28,020 zero and B equals nine. 22 00:02:28,350 --> 00:02:35,550 It means the generated random numbers will be in this range from zero to nine. 23 00:02:36,150 --> 00:02:40,500 Let's keep registering other objects. 24 00:02:41,040 --> 00:02:51,660 The next one individual those dots init repeat to generate random individuals. 25 00:02:52,170 --> 00:02:55,560 Greater thoughts individual. 26 00:02:56,130 --> 00:02:59,770 Let's call it bluebox dots. 27 00:03:00,650 --> 00:03:04,260 A data are aeons the same. 28 00:03:04,260 --> 00:03:17,430 We defined the year in this last line of code and now an equals 12 because there are 12 options, as 29 00:03:17,430 --> 00:03:18,750 we can see here. 30 00:03:19,410 --> 00:03:24,870 The next step is to register the population. 31 00:03:25,440 --> 00:03:29,300 Let's type population again. 32 00:03:29,880 --> 00:03:33,660 Those in need repeats. 33 00:03:35,080 --> 00:03:45,130 The population will be in the four of a list and it will be a population of individuals, according 34 00:03:45,130 --> 00:03:48,640 to the name given in these line here. 35 00:03:49,000 --> 00:03:58,750 Now we needs to register the order parameter, how we are going to evaluate the algorithm. 36 00:03:59,080 --> 00:04:03,220 Let's just type fitness function deep. 37 00:04:03,490 --> 00:04:13,570 The function we created last lecture that returns the total price of the flights, the next one, the 38 00:04:13,570 --> 00:04:17,260 box registered mate. 39 00:04:17,620 --> 00:04:30,310 The way we will apply the cross over again, we will use cross over one point, the next one, how we 40 00:04:30,310 --> 00:04:35,170 are going to perform the mutation new bait. 41 00:04:35,970 --> 00:04:45,250 Those but most flip bits in this case, here we can just change the values. 42 00:04:45,520 --> 00:04:53,530 For example, if it is three, maybe we can change use to seven two eight two nine. 43 00:04:53,920 --> 00:05:02,110 As we have more options, we can change to any integer value in the range from 029. 44 00:05:02,410 --> 00:05:10,780 And let's specify here the probability of mutation, for example, only one percent. 45 00:05:11,530 --> 00:05:20,350 The next one to box not register how we are going to select the best individuals. 46 00:05:20,770 --> 00:05:33,670 And before we were using this sale roulette, however, this method is used only for maximization problems. 47 00:05:34,030 --> 00:05:39,910 As in this particular case study, we have a minimization problem. 48 00:05:40,150 --> 00:05:49,870 We need to use order type of selection and there is one very famous, which is called star ornaments. 49 00:05:50,230 --> 00:05:54,280 We will use these authors selection technique. 50 00:05:54,700 --> 00:06:07,750 Then let's create the population equals tool box thought population and equals five hundred. 51 00:06:08,080 --> 00:06:15,790 This is the population size, as this problem has more data than the transplants of the products. 52 00:06:16,030 --> 00:06:23,710 We heeds a higher number here, and after executing the algorithm, you can choose other options. 53 00:06:24,370 --> 00:06:29,100 Let's create another variable crossover probability. 54 00:06:29,360 --> 00:06:43,030 I will set it to 70 percent and mutation probability to 30 percent, and the number of generations will 55 00:06:43,030 --> 00:06:45,220 be equal to 100. 56 00:06:45,460 --> 00:06:49,240 You can try other parameters and see the results. 57 00:06:49,690 --> 00:07:00,800 Finally, I will just based a code to hear the definition of the statistics and also these last bards. 58 00:07:00,820 --> 00:07:03,610 That's a short run, the algorithm. 59 00:07:03,910 --> 00:07:09,790 Again, we are calling it a simple and just correct here. 60 00:07:10,180 --> 00:07:12,520 Number of generations. 61 00:07:12,910 --> 00:07:14,800 Let's run this codes. 62 00:07:15,460 --> 00:07:24,280 There is one missing parameter when we are calling the tournaments, which is the dance size. 63 00:07:24,580 --> 00:07:28,390 Let's specify three, which is the default value. 64 00:07:28,900 --> 00:07:30,460 Run this code again. 65 00:07:30,670 --> 00:07:35,090 And finally, we can see the genetic algorithm running. 66 00:07:35,470 --> 00:07:46,570 We can see the first generation see that the minimum price was one thousand nine hundred and fifty six. 67 00:07:46,960 --> 00:07:56,260 And we can see that after the generations after running the genetic algorithms, we could find this 68 00:07:56,260 --> 00:07:57,430 value here. 69 00:07:57,670 --> 00:08:09,340 So the cost of all airline tickets for the round trip for all people would be this one here one thousand 70 00:08:09,430 --> 00:08:12,090 five hundred and sixty six. 71 00:08:12,460 --> 00:08:15,130 Again, this is a minimization problem. 72 00:08:15,400 --> 00:08:18,730 We can select the best individual. 73 00:08:19,180 --> 00:08:20,260 Less great. 74 00:08:20,740 --> 00:08:28,840 Best solution equals those dots sell best population. 75 00:08:29,380 --> 00:08:32,380 I want to select only one individual. 76 00:08:32,770 --> 00:08:34,420 Let's go through this. 77 00:08:34,460 --> 00:08:38,620 Solution and best solution. 78 00:08:39,010 --> 00:08:51,760 Unless friends, the individual and also lets brands individual got fitness here, we can see the flights 79 00:08:51,760 --> 00:08:53,410 that were chosen. 80 00:08:53,890 --> 00:08:56,710 And you see it in a bad way. 81 00:08:56,740 --> 00:09:04,480 We can call brilliance Kegel and let's sans as parameter the individual. 82 00:09:04,810 --> 00:09:06,460 Let's run this code. 83 00:09:06,790 --> 00:09:15,490 You can see here that we have a bad visualization, and it happened because they change the dysfunctional 84 00:09:15,490 --> 00:09:16,630 brain schedule. 85 00:09:16,960 --> 00:09:23,260 I just bought some additional spaces here should be more organized. 86 00:09:23,620 --> 00:09:33,610 We can see the ship flights from Lisbon to Rome and also the cheapest flight from Rome to Liverpool. 87 00:09:33,880 --> 00:09:44,830 And this value here is this sum of all the prices from this series to Rome in this column here and from 88 00:09:44,830 --> 00:09:49,810 Rome to each one of the other cities in this column here. 89 00:09:50,110 --> 00:09:57,640 And again, you can run this code a couple of times to get different results. 90 00:09:58,120 --> 00:10:01,660 Let's check what was the best value. 91 00:10:02,110 --> 00:10:09,790 See that we have exactly the same as before, and in the next lecture, we are going to solve the same 92 00:10:09,790 --> 00:10:13,030 problem using Melrose Library. 93 00:10:13,300 --> 00:10:13,960 See you there!