Claim Missing Document
Check
Articles

Found 2 Documents
Search
Journal : International Conference on Digital Advanced Tourism, Management, and Technology

Can Google Trends(GT) be used to predict tourist arrivals?: FB Prophet Machine Learning(ML) for Predicting Tourist Arrivals Indra Gunawan; Dwi Purnomo Putro; Adhika Pramita Widyassari 
International Conference On Digital Advanced Tourism Management And Technology Vol. 1 No. 1 (2023): International Conference on Digital Advanced Tourism, Management, and Technolog
Publisher : Sekolah Tinggi Ilmu Ekonomi Pariwisata Indonesia Semarang

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.56910/ictmt.v1i1.57

Abstract

The big problem in tourism is how to provide appropriate preparations to serve tourists so that when the tourist season is low, resources can be saved and when the tourist season is busy, all resources can be provided effectively. Machine learning is a derivative branch of artificial intelligence, one of whose capabilities can be used to carry out data/dataset-based forecasting. This research uses a dataset obtained from GT from 2013-2023 with several keywords combining city names and tourist destination names in Yogyakarta Indonesia, then it will be compared with a dataset of tourist arrivals in the city of Yogyakarta obtained from the Central Statistics Agency. The Machine Learning model that will be used is Prophet Facebook.. This model uses a Bayessian as a backend algorithm. The results obtained from this research are that GTs can be used to predict tourist arrivals with some tweaks on the dataset. However, to get accurate results, various combinations of keywords are needed for the desired destination, and it is recommended to add some column namely max and mean to the dataset to prevent insufficiency of data of some keywords that make prediction result bad. In this research it can be concluded that the use of an additional max column can increase the COERR, MAPE and R2 values. Meanwhile, we found that the GT dataset can be used for forecasting best in time periods under 200 days. Also we found that using the GT dataset alone produces unstable COERR, MAPE and R2 values. Another finding is that the GT dataset that uses the YouTube filter is only suitable for use in Indonesia for the time period above 2018 considering that Indonesian people's access to YouTube has increased massively over that year and tends to decrease below that year. However, the trend shows that the use of searches on YouTube after 2018 tends to increase drastically, beating searches on the Google web.
Collaboration of Progressive Web App (PWA) And Firebase Cloud Messaging (FCM) for Optimal Performance Mailing Software Dwi Purnomo Putro; Adhika Pramita Widyassari; Dea Salsabilla 
International Conference On Digital Advanced Tourism Management And Technology Vol. 1 No. 2 (2023): International Conference on Digital Advanced Tourism, Management, and Technolog
Publisher : Sekolah Tinggi Ilmu Ekonomi Pariwisata Indonesia Semarang

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.56910/ictmt.v1i2.126

Abstract

The problem of correspondence cannot be separated from the ease, accuracy and speed of the processing process. In research from Riswandi Ishak and H. Trizaka, they proposed report management software as well as notification of disposition and monitoring of correspondence as a solution to correspondence problems. However, there are shortcomings in the notification process for the disposition of correspondence, which requires the software to be actively open. In research, P. Dwi proposed a notification solution with Firebase Cloud Messaging (FCM), so that it can send notifications as long as the browser is active and connected to the internet, even without opening the software. There is a problem currently when the software becomes unstable when the internet connection is bad or offline. Progressive Web Apps (PWA) offers the concept of web-based application development by implementing browser technology such as service workers and app manifests. PWA is capable of displaying pages offline but cannot save, change, or delete data in the database. The test results of this research used Lighthouse and showed an average score of 100 on the PWA criteria, 85 on the performance criteria, 97 on the accessibility criteria, and 100 on the best practices criteria. Additional results obtained by implementing PWA mean page loading times are 26.6% faster with cache and service workers. The PWA and FCM concepts provide the best experience in using Mailing Software even with minimal internet connection or offline. This strategy was chosen to still get a fast response when running the mail processing software.