Optimalisasi penggunaan baterai, termasuk baterai VRLA yang sering digunakan untuk keperluan penyimpanan energi dalam jumlah besar dengan harga yang rendah, biasa diupayakan dengan menyelenggarakan Sistem Manajemen Baterai (SMB). Untuk melakukan SMB, dibutuhkan informasi tentang Kondisi Muatan (KM) dan Kondisi Kesehatan (KK) baterai. KM didefinisikan sebagai rasio kapasitas sisa baterai saat ini dengan kapasitas baterai sebelum pengosongan, sedangkan KK baterai ialah rasio antara kapasitas penuh terukur terhadap kapasitas nominalnya pada saat baterai masih dalam keadaan baru beroperasi. Yang menjadi permasalahan, kedua informasi tersebut tidak dapat diukur secara langsung. Estimasi KM dan KK dapat dilakukan secara tidak langsung dengan menggunakan parameter-parameter yang mudah diukur, terutama tegangan dan arus pada terminal baterai. Makalah ini menggunakan metode Coulomb Counting (CC) dan Regresi Vektor Pendukung (RVP) untuk mengestimasi KM dan KK baterai VRLA yang digunakan sebagai energi cadangan sistem nanogrid yang ada di laboratorium. Makalah ini menggunakan modul mesin pembelajaran Python yang memungkinkan implementasi RVP dengan berbagai jenis kernel, di antaranya kernel linear, kernel polinomial, dan kernel RBF. Pengujian yang dilakukan menggunakan modul grid search menunjukkan bahwa kinerja terbaik diperoleh ketika menggunakan kernel RBF.

The global demand for beauty products continues to grow due to raised public awareness of applying cosmetics, with a 1.45 % to 3.34 % growth annually. Unfortunately, the COVID-19 outbreak broke out globally in December 2019, affecting face-to-face businesses such as the beauty industry falling until –7.11 % in 2020. This study aims to analyze the impact of the COVID-19 outbreak on Indonesia’s beauty industry and the shift in the beauty consumer segment during the pandemic. This study adopts the react-cope-adapt (RCA) framework to construct the COVID-19 pandemic periodization in Indonesia. The correlation analysis was used to investigate the impact of the COVID-19 pandemic on the beauty industry. In addition, clustering techniques were employed to identify hidden consumer segments and product preferences throughout the COVID-19 outbreak. The study shows that COVID-19 cases positively impact beauty company’s sales during the reacting phase. A strong negative relationship between COVID-19 and company revenue was observed in the coping phase. In the adapt phase, the negative impact of COVID-19 on the company’s sales has decreased. Our finding also confirms the shift in consumer buying behavior during the pandemic. Consumers prefer to buy cosmetics products online than offline during the reaction phase. In the coping phase, consumers slowly begin to purchase in-store. Finally, consumers return to buying cosmetics offline in the adapting phase, similar to before the pandemic. The clustering results showed three hidden consumer segments: the loyal consumer segment, the impulsive consumer segment, and the compulsive consumer segment. In addition, during the pandemic, consumers prefer to buy skincare products over make-up products since government policies forced people to stay, work, and study at home. Our study has theoretical and practical implications. Theoretically, our results support the usefulness of the RCA model and clustering techniques in analyzing the change in consumer buying behavior during a time of crisis, such COVID-19 pandemic. Practically, beauty industries can anticipate this shift by accelerating the digital business transformation and focusing on the most preferred product to sustain their business