Matriks adalah salah satu materi matematika yang dianggap sulit. Materi matriks dianggap sulit utamanya karena banyaknya perhitungan yang harus dilakukan. Karena alasan itu pula, mencari kesalahan perhitungan juga sulit, sehingga membuat proses pembelajaran yang sebagian besar adalah pengerjaan latihan soal semakin sulit. Untuk mengatasi itu, dikembangkanlah suatu aplikasi penghitung matriks yang dapat menampilkan langkah-langkah perhitungannya. Pengembangan aplikasi ini menggunakan model prototipe yang memiliki lima fitur utama, yaitu pembantu operasi baris elementer, penghitung operasi matriks, penghitung bentuk-bentuk matriks, penghitung determinan matriks dengan ekspansi Laplace, dan penyelesaian persamaan linear dalam bentuk matriks dengan aturan Cramer. Pengujian aplikasi dilakukan dengan metode black box testing untuk menguji fungsionalitasnya yang menunjukkan bahwa setiap fitur aplikasi berfungsi dengan baik. Akan tetapi, aplikasi ini belum diuji efektivitasnya, sehingga masih dianggap prototipe.
COVID-19 is a dangerous illness because it spreads quickly and easily. Vaccines are already available but the pandemic isn’t likely to end soon. Forecasting is hoped to help handle the pandemic. Deep learning, specially LSTM, has been used to forecast COVID-19 case count in some regions. However, deep learning models generally need a lot of training data while COVID-19 daily data are scarce. However, COVID-19 pandemic happens in many regions. This research aims to use joint learning with data from other regions to improve model performance with fewer data and to use the model to forecast until 9 months since the date of last data taken. Joint learning was done by making models share some parts and training the models together. To overcome the different data scale and pandemic age in the regions, the data was first transformed into discrete SIRD variables and was evaluated using RMSSE. Joint learning failed to improve the model performance in this research. The proposed model performance was signficantly better than ARIMA-SIRD and SIRD model but wasn’t better than normal encoder-decoder LSTM. The models only reached RMSSE below one occasionally. Additionally, it was found that doing joint learning with all regions without selecting them by clustering can make the model performance worse instead. It was also found that RMSSE is too sensitive to a mostly stagnant time-series due to its division by the error of one-step naïve forecast.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.