Update Artikelilmiah: 55125

DINI YUNIA PUTRI

NIM

Judul Artikel

Abstrak (Bhs. Indonesia)

Stroberi (Fragaria × ananassa) merupakan buah yang kaya akan kandungan gizi dan banyak diminati konsumen. Namun, proses sortasinya di lapangan masih didasarkan pada karakteristik eksternal, seperti warna, ukuran, dan kondisi permukaan buah. Padahal, kualitas internal seperti tingkat kemanisan dan keasaman lebih menentukan penerimaan konsumen. Pengujian kualitas internal umumnya masih bersifat destruktif, menggunakan peralatan laboratorium canggih, serta memerlukan biaya relatif tinggi. Oleh karena itu, diperlukan metode non-destruktif yang cepat dan efisien dalam mengestimasi kualitas buah. Spektrometer UV–Vis–NIR berbasis sensor AS7265x memiliki potensi karena mampu memperoleh informasi spektral yang berkaitan dengan karakteristik fisikokimia buah. Penelitian ini juga mengintegrasikan penilaian sensori berupa tingkat kemanisan untuk mengkaji hubungan respons spektral dengan kualitas yang dirasakan konsumen. Penelitian ini bertujuan untuk: 1) menilai kinerja spektrometer UV–Vis–NIR berbasis sensor AS7265x dalam mengestimasi kualitas fisikokimia dan skorsensori stroberi, 2) membangun model machine learning berbasis data spektral menggunakan algoritma Simple Linear Regression (SLR), Multiple Linear Regression (MLR), dan Artificial Neural Network (ANN), serta 3) menentukan model machine learning dengan kinerja terbaik berdasarkan MAPE, RMSE, dan R2.

Penelitian ini dilaksanakan pada 11 November 2025 – 25 Februari 2026 di Laboratorium Teknologi Pertanian, Fakultas Pertanian, Universitas Jenderal Soedirman. Sampel penelitian terdiri atas 200 sampel stroberi yang berasal dari pasar tradisional dan pasar modern, masing-masing sebanyak 100 sampel. Setiap sampel terdiri atas lima buah stroberi yang digunakan untuk pengukuran spektrum, analisis fisikokimia, dan uji sensori dengan skala 1 (sangat tidak manis) hingga 5 (sangat manis). Pengukuran spektrum dilakukan menggunakan spektrometer berbasis sensor AS7265x yang memiliki rentang panjang gelombang 410 – 940 nm. Parameter yang diamati meliputi TPT, kadar air, vitamin C, kadar gula, serta sensori kemanisan. Data spektral diolah dan dimodelkan menggunakan algoritma SLR, MLR, dan ANN dengan metode 5-fold cross validation. Kinerja model dievaluasi menggunakan nilai MAPE, RMSE, dan R².

Hasil penelitian menunjukkan bahwa spektrometer UV–Vis–NIR berbasis sensor AS7265x mampu mengestimasi kualitas fisikokimia dan skor sensori stroberi secara nondestruktif. Ketiga algoritma berhasil membangun model prediksi, dengan ANN memberikan kinerja terbaik pada hampir keseluruhan parameter. Secara keseluruhan, model menghasilkan nilai MAPE sebesar 0,690 – 29,079%, RMSE sebesar 0,407 – 10,393, dan R² sebesar 0,064 – 0,589. Hasil menunjukkan bahwa kemampuan prediksi berbeda pada setiap parameter, dengan MAPE dan RMSE yang menunjukkan tingkat kesalahan prediksi yang bervariasi, sementara nilai R² yang relatif rendah menunjukkan bahwa kemampuan model dalam menjelaskan hubungan antara data spektral dan kualitas stroberi masih terbatas.

Abtrak (Bhs. Inggris)

Strawberry (Fragaria × ananassa) is a fruit rich in nutrients and widely favored by consumers. However, fruit sorting in the field is still primarily based on external characteristics, such as color, size, and surface condition, even though internal quality attributes, including sweetness and acidity, play a more important role in consumer acceptance. Internal quality assessment is generally destructive, requires laboratory equipment, and involves relatively high costs analysis. Therefore, a rapid and efficient non-destructive method for estimating fruit quality is needed. A UV–Vis–NIR spectrometer based on the AS7265x sensor has considerable potential because it can acquire spectral information related to the physicochemical characteristics of fruit. This study also integrated sensory evaluation of sweetness to investigate the relationship between spectral responses and consumer-perceived quality. The objectives of this study were to: (1) evaluate the performance of the AS7265x based UV–Vis–NIR spectrometer in estimating the physicochemical quality and sensory scores of strawberries, (2) develop spectral data based machine learning models using Simple Linear Regression (SLR), Multiple Linear Regression (MLR), and Artificial Neural Network (ANN) algorithms, and (3) determine the best performing machine learning model based on MAPE, RMSE, and R².

This research was conducted from 11 November 2025 – 25 February 2026 at the Agricultural Technology Laboratory, Faculty of Agriculture, Jenderal Soedirman University. A total of 200 strawberry samples were collected from traditional and modern markets, with 100 samples from each market type. Each sample consisted of five strawberries used for spectral measurements, physicochemical analysis, and sensory sweetness evaluation using a five-point scale ranging from 1 (very not sweet) to 5 (very
sweet). Spectral measurements were performed using an AS7265x spectrometer covering a wavelength range of 410 – 940 nm. The observed parameters included total soluble solids (TSS), moisture content, vitamin C, sugar content, and sensory scores. The spectral data were modeled using SLR, MLR, and ANN algorithms with a 5-fold cross-validation approach. Model performance was evaluated using MAPE, RMSE, and R².

The results showed that the UV–Vis–NIR spectrometer based on the AS7265x sensor was capable of estimating the physicochemical quality and sensory scores of strawberries non-destructively. The three machine learning algorithms successfully developed prediction models, with ANN demonstrating the best performance for nearly all parameters. Overall, the models produced MAPE values ranging from 0.690 – 29.079%, RMSE values ranging from 0.407 – 10.393, and R² values ranging from 0.064 – 0.589. These results indicate that prediction performance varied across parameters, with the MAPE and RMSE values reflecting varying levels of prediction error, while the relatively low R² values indicate that the models had limited ability to explain the relationship between the spectral data and strawberry quality.

Kata kunci

Pembimbing 1

Pembimbing 2

Pembimbing 3

Tahun

Jumlah Halaman