Fausi, Ria Risti and STP., M.SC, Dr. Mochamad Bagus Hermanto and STP, M. Si, Zaqlul Iqbal (2022) Pembangunan Model Prediksi Mutu Kopi Arabika (Coffea arabica) Sangrai Berbasis Spektroskopi Vis-NIR dan Analisis Kemometrika. Sarjana thesis, Universitas Brawijaya.
Abstract
Sebagai upaya untuk menjaga kualitas biji kopi roasting, faktor fisiko-kimia penting untuk diperhatikan. Jaminan kualitas secara fisik seperti mutu sensori dapat dilakukan melalui uji cupping test sedangkan kandungan senyawa kimiawi melalui uji laboratorium. Namun, kedua metode tersebut membutuhkan waktu lama, serta akan menghabiskan banyak biaya. Oleh karena itu, dibutuhkan metode yang lebih cepat, sederhana dan murah, seperti penggunaan metode spektroskopi Vis-NIR. Penelitian ini bertujuan untuk membangun model prediksi mutu sensori, TDS, brix dan kandungan kafein pada kopi Arabika (coffea arabica) sangrai menggunakan instrumen deteksi cepat berbasis spektroskopi Vis-NIR dengan analisis kemometrik. Penelitian ini dilaksanakan mulai dari bulan Juni sampai November 2022 di Laboratorium Kimia Dasar, Laboratorium DDM, Laboratorium Mekatronika dan Robotika FTP UB, Kopi Wonosantri Malang, serta Sensoflavo Malang. Tahapan penelitian terdiri dari persiapan sampel biji kopi roasting pada level medium suhu 198°C dengan waktu roasting yaitu 6 ,10, dan 14 menit. Kedua yaitu akuisisi data yang terdiri dari data spektra menggunakan instrumen deteksi berbasiskan spektrofotometer Vernier Go Direct Spektro Vis, data kafein dengan uji laboratorium, data TDS, brix dan cupping test menggunakan jasa panelis. Data yang diperoleh selanjutnya diberikan perlakuan preprocessing Moving Average (MA), SNV, MSC. Kemudian diolah menggunakan metode PLS-DA untuk klasifikasi dan PLSR untuk membangun model prediksi dengan software python. Hasil analisis PLS-DA menunjukkan bahwa dengan perlakuan MA dan MSC mampu mengklasifikasikan sampel dengan baik berdasarkan perbedaan waktu roasting. Diperoleh nilai accuracy, precision dan sensitivity 83%, 88%, 86% untuk model testing. Kemudian, hasil analisis PLSR menunjukkan bahwa hanya beberapa parameter mutu internal yang memenuhi parameter R2 lebih dari 0.7, nilai error kecil dan RPD>1.5 yaitu mutu sensori seperti acidity, body, balance dan total score. Sedangkan parameter lainnya hanya dapat memenuhi syarat parameter R2 dan error saja serta masih belum mampu memenuhi syarat parameter RPD.
English Abstract
To maintain the quality of roasted coffee beans, Physico-chemical factors are important to pay attention to. Assurance of physical quality such as sensory quality can be carried out through a cupping test while the content of chemical compounds through laboratory tests. However, both methods are time-consuming and costly. Therefore, a faster, simpler, and cheaper method is needed, such as using the Vis-NIR spectroscopy method. This study aims to build a predictive model for sensory quality, TDS, brix, and caffeine content in roasted Arabica coffee (Coffea arabica) using a fast detection instrument based on Vis-NIR spectroscopy with chemometric analysis. This research was carried out from June to November 2022 at the Basic Chemistry Laboratory, DDM Laboratory, Mechatronics and Robotics Laboratory of FTP UB, Kopi Wonosantri Malang and Sensoflavo Malang. The research phase consisted of preparing coffee bean samples for roasting at a medium temperature level of 198°C with 6, 10, and 14 minutes roasting times. The second is data acquisition consisting of spectral data using a detection instrument based on the Vernier Go Direct Spektro Vis spectrophotometer, caffeine data with laboratory tests, TDS data, brix, and cupping tests using the services of panelists. The data obtained is then given preprocessing treatment such as Moving Average (MA), SNV, and MSC. Then it is processed using the PLS-DA method for classification and PLSR for building prediction models with Python software. The results of the PLS-DA analysis showed that the MA and MSC treatments were able to classify samples properly based on differences in roasting time. Accuracy, precision and sensitivity values obtained are 83%, 88%, and 86% for model testing. Then, the results of the PLSR analysis showed that only a few internal quality parameters met the R2 parameter of more than 0.7, small error values, and RPD> 1.5, namely sensory qualities such as acidity, body, balance, and total score. While the other parameters can only meet the requirements for parameter R2, error and are still unable to meet the requirements for the RPD parameter.
Item Type: | Thesis (Sarjana) |
---|---|
Identification Number: | 0522100508 |
Uncontrolled Keywords: | Biji kopi Roasting, Kemometrika, Spektrofotometer Vernier, Spektroskopi Vis-NIR, Chemometrics, Roasted coffee beans, Vernier Spectrophotometer, Vis-NIR spectroscopy. |
Subjects: | 600 Technology (Applied sciences) > 630 Agriculture and related technologies |
Divisions: | Fakultas Teknologi Pertanian > Keteknikan Pertanian |
Depositing User: | soegeng sugeng |
Date Deposited: | 22 May 2023 01:17 |
Last Modified: | 22 May 2023 01:17 |
URI: | http://repository.ub.ac.id/id/eprint/199938 |
Text (DALAM MASA EMBARGO)
Ria Risti Fausi.pdf Restricted to Registered users only until 31 December 2024. Download (3MB) |
Actions (login required)
View Item |