Analysis and Design of a Route Recommendation System and Bicycle Rental Fees at Tourist Destinations with Genetic Algorithms

− The tourism sector is one of the country's most significant sources of income, especially in Indonesia. This is inseparable from many visitors or foreign tourists who come to Indonesia. Bandung is a tourist destination that is famous for its natural beauty and traditional culture, one example is in Desa Alamendah which has several tourist destinations, namely agro-tourism, nature tourism, and cultural tourism. However, access that can be passed from every tourist spot in Desa Alamendah can only be given on foot or by bicycle. Responding to this matter, bicycle facilities as a means of rotation are needed to access tourist attractions in Desa Alamendah. Based on these problems, this study aims to create a recommendation system for tourist attractions and bicycle rentals using an application in Desa Alamendah, Bandung. Using the Genetic Algorithm, an algorithm that can solve multi-objective problems can be applied to a tourist spot recommendation system in Desa Alamendah. In this system, the Genetic Algorithm is also used to determine fares based on the total distance traveled, so that bicycle rentals are more efficient. The study combines Google Maps as an appropriate route selection based on the recommendation of tourist destination points that tourists can visit using Genetic Algorithms, so that bicycle rentals are more efficient. This is study combines Google Maps as an appropriate route selection based on the recommendation of tourist destination points that tourists can visit using Genetic Algorithms. So that bicycle rentals are more efficient. The study combines Google Maps as an appropriate route selection based on the recommendation of tourist destination points that tourists can visit using Genetic Algorithms. Based on the results of tests that have been carried out in the process of forming tourist attractions recommendations with a Genetic Algorithm that using mutation probability 1.0 and Crossover Probability 0.6, can produce a Mobile Application with a Genetic Algorithm as a tourist spot recommendation system in Desa Alamendah. In addition, it also can provide recommendations, namely displaying tourist points, route using Google Maps. The rental fee is based on the total distance traveled. A study named "An optimization model for product returns using genetic algorithms and artificial immune system" was undertaken by Ali Diabat, Devika Kannan, Mathiyazhagan Kaliyan, and Davor Svetinovic (2013). This study creates a mathematical model that connects clients, initial collection places, and control centers in order to optimize reverse logistics. However, network design for used products is time-consuming and complex. The number and location of starting collection points and return centers, as well as the maximum waiting time (collection frequency) for the sum of tiny volumes of returned items, must be determined using a mixed-integer non-linear programming (MINLP) model developed for this study. There are two techniques to finding a solution[7].


INTRODUCTION
The tourism sector is a sector that plays a vital role in increasing income. Sustainable tourism not only considers its economic, social, and environmental impacts but must also address the needs of tourists, industry, the environment, and the local population. [1]. Bandung is a tourist destination that is famous for its natural beauty and traditional culture, one example is in Desa Alamendah, which has several tourist destinations, namely agrotourism, nature tourism, and cultural tourism. However, access that can be passed from every tourist spot in Desa Alamendah can only be done on foot or by bicycle. Responding to this matter, bicycle facilities as a means of ratio are needed to access tourist attractions in Desa Alamendah.
Bicycles offer an advantage over other vehicles likely to produce pollution; besides, other vehicles often cause traffic jams in tourist areas [2]. Therefore, a bicycle rental system is needed at tourist attractions. To facilitate access to transportation, for tourists to be able to explore all tourist attractions.
Study [3]proposed a Genetic Algorithm for a bicycle sharing system for tourism. Their research aims to improve bicycle-sharing services, thanks to the Mobile Tourist Guide (MTG), which can support bicycle renters to recommend exciting places to visit. Based on this research, it takes a simulation of destination point recommendations, and the author will build the application by combining it with Google Maps. In a study [4], the Genetic Algorithm can solve multi-objective problems, applying it to a culinary tourism recommendation system. The crossover method with one cut point and mutation with gene shift in the Genetic Algorithm can solve the problem. Each destination visited also has time and load constraints. Close the place. Based on this research, it takes a simulation of destination point recommendations, and the author will build the application by combining it with Google Maps. As a recommendation system to tourist destination points, with the development of tourist spot recommendations based on real-time time, the application will automatically recommend tourist attractions at the right time according to the schedule of opening and closing hours.
In Desa Alamendah, there are various kinds of tourist attractions, sometimes making it difficult for tourists to choose which tourist attractions they will visit. The Genetic Algorithm will recommend tourists to visit tourist attractions in Desa Alamendah by knowing which tourist attractions are open or closed when tourists are at the location. This is implemented into an application on a smartphone [5], which can detect where the area of tourists is [6]so that you can find out all the tourist points that have been divided according to opening and closing hours. Therefore, it is hoped that tourists will get an easy route to visit tourist attractions in Desa Alamendah.
In designing the application system, it uses Genetic Algorithms to recommend tourist destination points that tourists can pass. The total places traveled by tourists will also be calculated using a Genetic Algorithm and route results or directions displayed using Google Maps.
A study named "An optimization model for product returns using genetic algorithms and artificial immune system" was undertaken by Ali Diabat, Devika Kannan, Mathiyazhagan Kaliyan, and Davor Svetinovic (2013). This study creates a mathematical model that connects clients, initial collection places, and control centers in order to optimize reverse logistics. However, network design for used products is time-consuming and complex. The number and location of starting collection points and return centers, as well as the maximum waiting time (collection frequency) for the sum of tiny volumes of returned items, must be determined using a mixed-integer non-linear programming (MINLP) model developed for this study. There are two techniques to finding a solution [7].
In the following research by Farhad Ghassemi Tari Zahra Hashemi (2016) conducted a study entitled "A priority-based genetic algorithm for nonlinear transportation costs problems." In this study, the problem of allocating goods delivery vehicles from manufacturing companies to goods storage depots. Each depot has a certain number of orders, and transportation costs consist of fixed and variable transportation costs. The goal is to assign the right type and number of vehicles to each depot lane to minimize transportation costs. To succeed in today's rapidly evolving business competition, companies need to design and build logistics models that are productive and cost-effective. Therefore, the author uses a genetic algorithm to overcome the distribution of vehicles that have been determined [8]. A new real-time personalized route recommendation system, as presented in this study, is needed to solve the problem and to provide route recommendation services for the increasing number of self-driving tourists in tourist attractions by Long Liu, Jin Xu, Stephen Shaoyi Liao, Huaping Chen, who entitled "A real-time personalized route recommendation system for self-drive tourists based on vehicle to vehicle communication" Finally, with real-time queue information, a recommended path was developed based on the interests and preferences of comparable travelers in the same group [9].
Based on research conducted by Kevin Krisnandi, Halim Agung (2017) conducted a study entitled "Implementation of Genetic Algorithms to Predict Construction Project Work Time and Costs." The problems faced by construction companies have stages in working on a construction project. To obtain accurate results, an algorithm is needed that is used in the Application of Prediction of Time and Cost of Work on Construction Projects, namely a Genetic Algorithm that helps construction companies predict the time and costs required to complete a construction project. In addition to predicting this application can also provide solutions to the foreman if there are problems in working on a construction project so that the project can run according to the predicted time. . To obtain accurate results, an algorithm is needed that is used in the Application of Prediction of Time and Cost of Work on Construction Projects, namely a Genetic Algorithm that helps construction companies predict the time and costs required to complete a construction project [10]. Genetic algorithms based on a variety of crossover operators have been published in the literature to solve this problem; selecting an effective crossover operator can result in a successful GA. For resolving the MTSP entitled "Genetic Algorithms for the Multiple Travelling Salesman Problem," Maha Ata Al-Furhud and Zakir Hussain Ahmed propose employing sequential constructive crossover (SCX), adaptive SCX, greedy SCX, reverse greedy SCX, and comprehensive SCX to produce simple and effective GAs. Because this paper offers genetic algorithms (GAs) to solve the problem, we conduct a literature review on GAs, focusing on crossover operators [11].
In another research on Genetic Algorithms that have been carried out by Herdiesel Santoso Rachmad Sanuri(2019) conducted a study entitled "Implementation of Genetic Algorithms and Google Maps APIs in Solving Traveling Salesman Problems with Time Window (TSP-TW) in Scheduling Travel Routes in the Marketing Division of STMIK El Rahma." In this study, the test scenarios were testing the optimal number of generations, testing the optimal number of populations, testing the combination of crossover probability (Pc) and mutation probability (Pm), and testing the consistency of the solutions produced by the Genetic Algorithm. Recommendations are given by considering the distance and travel time Symmetrical and asymmetrical and traffic density obtained from Google Maps using Google Maps. The results of the study are routes with costs [12].
Then as for research conducted by Handy Eka Putra Anwar, Deviani Titi Nautami, Dhina Puspasari Wijaya (2016), conducted a study entitled "Prototype of Search System Recommendations for Local Craft Locations in Yogyakarta Using Traveling Salesman Problem (TSP) With Genetic Algorithm." In this study, TSP with genetic algorithms helps determine what and where the locations of local crafts in Yogyakarta are in the grip of their smartphone by utilizing the Global Positioning System (GPS) to detect where they are located that can be determined at any point. -local craft points that have been categorized according to the types of local crafts in Yogyakarta, the genetic algorithm aims to find the individual with the highest fitness value [13].
Another research conducted by Cokorda Pramartha and Hendra Saputra (2020) conducted a study entitled "Recommendations for Web-Based Travel Routes Using Genetic Algorithms." In this study, based on the problem, the sites on the internet only display information and cannot respond and interact directly with the user's particular needs such as a travel agenda, the problem has been resolved previously), the prototype built by the author has been able to provide recommendations according to user needs. The system that was built still has a weakness. Namely, the system will produce too many recommendations (less than optimal because not all of them are appropriate) [14]. Multi-trip is a method of resolving the VRPTW that focuses on trip scheduling to achieve the best results. To overcome the problem, a genetic algorithm is combined with a simulated annealing algorithm in this study. By Amalia Kartika Ariyani, Wayan Firdaus Mahmudy, Yusuf Priyo Anggodo. Who entitled "Hybrid Genetic Algorithms and Simulated Annealing for Multi-trip Vehicle Routing Problem with Time Windows". In

Methodology
The research methodology can be seen in Figure 1, the research stage as a series of processes from the beginning to the end of the research. The first process of designing a design system to explain the design of the application system used UML, namely Use Case Diagrams and Activity diagrams, illustrating the flow of the Genetic Algorithm on the Application system using Flowcharts. The following process discusses the implementation of the system at this stage, explaining all the functions, menus and icons in the built application. In the Implemented Genetic Algorithm determining the parameters of the Genetic Algorithm in this study and also explaining the flow of the Genetic Algorithm that is implemented into the application being built, In the final stage namely Test Result getting an output from the results of the Genetic Algorithm which is implemented from a predetermined test simulation.

Genetic Algorithm
A Genetic Algorithm is a branch of Evolutionary algorithms based on the theory of natural selection. Genetic Algorithms are commonly used to solve optimization problems [16]. Genetic Algorithm imitates the analogy of biological evolution to determine high-quality chromosomes or individuals in a potential area called a population. The population consists of several individuals where each individual represents a solution. Each population contains several chromosomes composed of several genes [17]. Using the Genetic Algorithm in this system recommends tourist destination points based on the right time to visit, then displays tourist routes obtained from Google Maps [18].

System Design
In this study, the method for recommending tourist attractions uses a Genetic Algorithm. In this design sub-chapter, UML is used as a model that aims to help explain the design of the application system. A Flowchart is used for modeling the flow of the Genetic Algorithm Application system. The UML used in this research is Use Case Diagrams and Activity Diagrams.

Use Case Diagram
Use case diagrams serve to describe the outline of the application, which is the relationship between tourists and the system. The following use case diagram used in this study can be seen in Figure 2.

Activity Diagram
The activity diagram describes the system's flow designed for the application of tourist attractions recommendations the following activity diagrams used in this study can be seen in Figure 2.

Genetic Algorithm Flowchart
The following is a flowchart according to Figure 4. that describes the process of the Genetic Algorithm used in the Application as a tourist spot recommendation system and calculates the total distance traveled by bicycle.

Figure 4. Genetic Algorithm Flowchart
It can be seen from the flowchart above first the system receives input, namely the user's starting point or initial location when tourists are at tourist attractions. In the second stage, initialize the Genetic Algorithm population to become a tourist spot. Each population will form an individual. The third stage is an evaluation of the Fitness Value, and then at the Fitness value evaluation stage, the selection is carried out the crossover. Until you get the results in directions from 1 tourist spot to another and the total distance between tourist attractions and rates. Smartphones in this application are used to receive data between Smartphones and APIs, using internet connections, APIs that include Latitude Longitude Data for Tourist Attractions. Genetic Algorithm is used to recommend paths to tourist attractions based on the opening and closing hours, also calculates bicycle rental rates based on the total distance traveled by tourists. Google Maps provides a map view of Desa Alamendah and displays the route to each tourist point.

Tourist Place Data
This research used 14 tourist destinations in Desa Alamendah, covering three parts: cultural tourism, nature tourism, and agrotourism. In accordance with table 1, tourist attractions that contains the coordinates of tourist attractions along with latitude and longitude values.

RESULTS AND DISCUSSION
In this research, an application is made that helps tourists choose tourist attractions that can be visited. Because the system will provide recommendations to tourist attractions within a relatively short distance, this application can be accessed via the Mobile Application using the Kotlin language as the programming language. Google Maps is used to receive and retrieve coordinates and provide route directions by displaying the route.

Implementation
In the implementation of this system, there are four display pages. Namely, Home View, Display My Location, Desa Alamendah Map Display, Display of Total Distance and Fares. Each of these display pages has a different function, and the icon for each page will be explained below. Figure 6 This is the initial view of the application.  Recommendations for tourist attractions using the Genetic Algorithm can be seen in Figure 8 there is a black line. It is a path that tourists can traverse. The path will disappear according to the opening hours of tourist attractions. It can be seen in Figure 8. The route appears taking from the schedule of tourist attractions that are open at 14.00 by recommending ten tourist spots. You can also see a list of tourist destinations in the list text display. By pressing the show icon can be seen in Figure 9. Figure 10 is the Display of Total Distance and Fares. The icon can be pressed with the function described below.

Figure 10. Display Total Distance and Fares
To find out the total distance and fares from all tourist attractions, tourists can press the green icon as shown in Figure 10. The system will display the total distance and fares from all tourist attractions.

Implemented Genetic Algorithm
The Genetic Algorithm used in this study consisted of Population Initialization, selection, crossover process, and mutation process, Generation Termination. The following from table 2. The parameters and stages of the Genetic Algorithm in the Desa Alamendah Tourist Place Application System.

Genetic Algorithm Parameters Value
Total Population 10 Mutation Probability 1.0 Crossover Probability 0.6

Population Initialization
The population Initialization Process takes several chromosomes, and each chromosome contains a gene. Gene is a representation of Tourist Places. The arrangement of the chromosomes is a different series of chromosomes; the gene formation is taken at random. The initial population in this study contained ten chromosomes, with as many

Selection
The selection process is the selection of a chromosome from the population used for the crossover process. The selection results are tested, comparing the distance between 2 tourist attractions. With the final result, the closest distance will be used as the parent.

Crossover
The crossover process aims to produce a new chromosome or child by comparing the new chromosome, which is better than the parent obtained from the selection process. Selected according to the shortest distance.

Mutation
The Mutation process aims to get a new chromosome or Child. The results of the new chromosome are obtained from the crossover results, the best value is selected.

Generation Termination
Generation termination is done after several iterations, and the Genetic Algorithm process will stop when it reaches the initial parameters, namely Mutation Probability 1.0 and Crossover Probability 0.6. The best chromosomes from the population can be seen in Table 4. The 10th chromosome is the best because it produces the shortest total distance and the lowest total rate compared to other chromosomes.

Test Results
The Genetic Algorithm results using two samples of tourist starting point locations can be seen in Table 5.   Table 5 takes two samples of the location of the tourist starting point, namely Jalan Raya Cibodas and Jalan Cikembang. The schedule of the tourist attractions is taken from Table 1. From the 2 locations, it can be seen that there are differences in the recommendations for each hour given by the Genetic Algorithm, and the total distance and total fare also differ in results, the output of the recommendations generated by the genetic algorithm can be seen in Figures 11 and 12.