CARS-GA-SVM Model for Accurate Soluble Solids Content Prediction in Apples via NIR Spectroscopy: Raw Data and Analysis Codes

This dataset contains all raw spectral measurements, preprocessing outputs, feature selection results, regression scripts and intelligent optimization codes to reproduce the soluble solids content (SSC) prediction model of apple fruit based on visible-near infrared spectroscopy in the submitted manuscript CARS-GA-SVM Model for Accurate Soluble Solids Content Prediction in Apples via NIR Spectroscopy. The dataset is divided into five folders: Raw Spectral Data: Original reflectance spectra of 744 apple samples collected in the range of 380–1100 nm, paired with reference SSC values measured by refractometer. Preprocessed Spectral Data: Spectral datasets processed by SNV, MSC, Savitzky–Golay smoothing, first/second derivatives and standardization. Dimensionality Reduction Results: Wavelength screening outputs of CARS, SPA, BOSS and GA feature selection algorithms, including regression coefficient curves and optimal variable subsets. Regression Models: MATLAB scripts for PLS and SVM baseline regression modeling. Model Optimization Files: Genetic Algorithm (GA), Particle Swarm Optimization (PSO) and Sparrow Search Algorithm (SSA) codes for SVM hyperparameter tuning, iteration records and optimal model performance indicators.