Pengembangan Back-end Dan Model Machine Learning Sistem Identifikasi Nanofosil Dan Fossil List Berbasis Web
RADITYA CAHYO ADHI, Dr.Eng. Silmi Fauziati, S.T., M.T.;Ir. Noor Akhmad Setiawan, S.T., M.T., Ph.D., IPM.
2023 | Skripsi | S1 TEKNOLOGI INFORMASIDalam penelitian terkait biostratigrafi dengan nanofosil, dalam melakukan identifikasi nanofosil, peneliti geologi biasanya perlu mengamati ciri-ciri nanofosil melalui mikroskop. Selain itu, peneliti juga perlu untuk membuat fossil list untuk mencari tahu perkiraan umur lapisan batuan. Proses tersebut dirasa sulit serta membutuhkan waktu yang lama. Di samping itu, agar hasil identifikasi akurat, diperlukan pengetahuan yang mendalam tentang nanofosil. Dalam suatu penelitian sebelumnya, telah dikembangkan suatu sistem pakar identifikasi spesies mikrofosil foraminifera berbasis web. Namun, tingkat akurasinya masih di bawah 85%. Lalu, dalam penelitian lain, telah dikembangkan sistem pelaporan otomatis chart fossil list mikrofosil planktonik foraminifera berbasis web. Namun, fossil list yang dibuat hanya dapat ditampilkan dalam web browser dan dicetak dalam bentuk fisik tanpa dapat diunduh sebagai file dalam berbagai format. Pada Capstone Project ini dikembangkan sistem web identifikasi nanofosil dengan machine learning dan pembuatan fossil list yang dapat diunduh dalam berbagai format bernama Nannomate untuk membantu menyelesaikan permasalahan tersebut. Machine learning identifikasi dikembangkan menggunakan bahasa pemrograman Python dan algoritma Random Forest, back-end server machine learning menggunakan bahasa pemrograman Python dan framework Flask, dan back-end server database menggunakan bahasa pemrograman PHP, framework Laravel, serta database MariaDB. Pengujian model machine learning menggunakan pengujian performa, pengujian API back-end menggunakan metode basis-path testing.
In research related to biostratigraphy with nannofossils, in conducting nannofossil identification, geological researchers usually need to observe the characteristics of nannofossils through a microscope. In addition, researchers also need to make a fossil list to find out the approximate age of the rock layer. The process is considered difficult and takes a long time. In addition, for the identification results to be accurate, in-depth knowledge of nannofossils is required. In a previous study, a web-based foraminifera microfossil species identification expert system was developed. However, the accuracy rate is still below 85%. Then, in another study, a web-based automatic reporting system for fossil list chart of planktonic foraminifera microfossils has been developed. However, the fossil list created can only be displayed in a web browser and printed in physical form without being able to be downloaded as a file in various formats. In this Capstone Project, a web system for identifying nannofossils with machine learning and creating a downloadable fossil list in various formats called Nannomate was developed to help solve the problem. The identification machine learning was developed using Python programming language and Random Forest algorithm, the machine learning server back-end using Python programming language and Flask framework, and the database server back-end using PHP programming language, Laravel framework, and MariaDB database. Machine learning model testing used performance testing, back-end API testing used basis-path testing method.
Kata Kunci : Web,Identifikasi,Fossil List,Nanofosil,Machine Learning
