Electronic Theses and Dissertation

Penelitian ini bertujuan untuk mengembangkan model prediksi kandungan nitrogen pada ecoenzim berbahan limbah organik menggunakan teknologi near infrared spectroscopy (nirs). analisis kandungan nitrogen secara konvensional umumnya memerlukan waktu yang relatif lama, bersifat destruktif, dan menggunakan bahan kimia, sehingga diperlukan metode alternatif yang lebih cepat dan efisien. sampel ecoenzim dibuat dari berbagai jenis limbah organik yang difermentasi selama 90 hari, kemudian dilakukan pengukuran spektrum nirs dan analisis kandungan nitrogen sebagai data referensi. data spektrum diolah menggunakan perangkat lunak the unscrambler dengan metode partial least squares (pls) serta beberapa perlakuan awal (pretreatment). kinerja model dievaluasi berdasarkan nilai koefisien korelasi (r), koefisien determinasi (r²), root mean square error of calibration (rmsec), dan residual predictive deviation (rpd). hasil penelitian menunjukkan bahwa model terbaik diperoleh menggunakan metode pls dengan pretreatment first derivative (d1), menghasilkan nilai r = 0,9481, r² = 0,8989, rmsec = 0,0028, dan rpd = 3,2142. hasil tersebut menunjukkan bahwa model memiliki tingkat akurasi yang baik dalam memprediksi kandungan nitrogen pada ecoenzim. dengan demikian, teknologi nirs berpotensi digunakan sebagai metode analisis kandungan nitrogen yang cepat, non-destruktif, dan efisien pada ecoenzim

This study aims to develop a prediction model for nitrogen content in eco-enzymes derived from organic waste using Near Infrared Spectroscopy (NIRS) technology. Conventional nitrogen analysis is generally time-consuming, destructive, and involves the use of chemicals; therefore, a faster and more efficient alternative method is required. Eco-enzyme samples were prepared from various types of organic waste fermented for 90 days, followed by NIRS spectral measurements and nitrogen content analysis to serve as reference data. The spectral data were processed using The Unscrambler software employing the Partial Least Squares (PLS) method and various pretreatment techniques. Model performance was evaluated based on the correlation coefficient (r), coefficient of determination (R²), Root Mean Square Error of Calibration (RMSEC), and Residual Predictive Deviation (RPD). The results indicate that the best model was obtained using the PLS method with First Derivative (D1) pretreatment, yielding values ​​of r = 0.9481, R² = 0.8989, RMSEC = 0.0028, and RPD = 3.2142. These results demonstrate that the model possesses a good level of accuracy in predicting nitrogen content in eco-enzymes. Thus, NIRS technology has the potential to serve as a rapid, non-destructive, and efficient method for analyzing nitrogen content in eco-enzymes.