Data analysis on near infrared spectroscopy as a part of technology adoption for cocoa farmer in Aceh Province, Indonesia

Agussabti; Rahmaddiansyah; Purwana Satriyo; Agus Arip Munawar

doi:10.1016/j.dib.2020.105251

Data analysis on near infrared spectroscopy as a part of technology adoption for cocoa farmer in Aceh Province, Indonesia

Data Brief. 2020 Feb 6:29:105251. doi: 10.1016/j.dib.2020.105251. eCollection 2020 Apr.

Authors

Agussabti¹, Rahmaddiansyah¹, Purwana Satriyo², Agus Arip Munawar^{2

3}

Affiliations

¹ Department of Agribusiness, Faculty of Agriculture, Syiah Kuala University, Banda Aceh, Indonesia.
² Department of Agricultural Engineering, Faculty of Agriculture, Syiah Kuala University, Banda Aceh, Indonesia.
³ Agricultural Mechanization Research Centre, Syiah Kuala University, Banda Aceh, Indonesia.

Abstract

Presented manuscript described data analysis on near infrared spectroscopy used as adopted and portable technology for cocoa farmers in Aceh Province, Indonesia. The near infrared spectroscopy (NIRS) assisted farmers in post-harvest handling especially for cocoa quality evaluation. This technology was used to determine moisture content (MC) and fat content (FC) of intact cocoa bean samples rapidly and simultaneously. Near infrared spectra data were acquired as absorbance spectrum in wavelength range from 1000 to 2500 nm with co-added of 32 scans for a total of 72 intact bulk cocoa bean samples. Spectra data can be used to predict MC and FC of intact cocoa beans by establishing prediction models and validate with actual MC and FC measured by means of standard laboratory procedures. Prediction performances were evaluated using several statistical indicators: coefficient correlation (r), coefficient of determination (R²), root mean square error (RMSE) and residual predictive deviation (RPD) index. Near infrared spectra data can be enhanced using spectra pre-treatment methods to improve prediction performances. Moreover, prediction models can be developed using principal component regression (PCR), partial least squares regression (PLSR) and other regression approaches. Ideal prediction models should have r and R² above 0.75, RPD index above 2.0 and RMSE lower than its standard deviation (SD). Dataset were available as raw MS Excel format and The Unscrambler files as *.unsb extension.

Keywords: Cocoa; NIRS; Post-harvest; Spectroscopy; Technology.