3D Printed Cuvette Sensor Data for Real-time Soil Nutrient Sensing

This dataset presents experimental measurements for soil macronutrient (N, P, and K) estimation using a custom-built visible transmission spectroscopy system. Ten soil sample solutions were prepared with varying quantities of de-ionized water, resulting in 100 distinct soil samples forming 100 different nutrient (N,P,K) concentrations. Each sample was analyzed using six multiplexed LEDs spanning the visible spectrum, selected to align with nutrient-specific absorbance regions: 450\u2013500 nm (N), 510\u2013615 nm (P), and 605\u2013630 nm (K). For each concentration level, intensity measurements were recorded over a 30-second interval, yielding approximately 1,000 data points per LED, and repeated six times to account for variability. The resulting dataset contains both raw sensor outputs and corresponding ground-truth nutrient concentrations obtained through laboratory analysis.This dataset is valuable for developing and benchmarking machine learning models for nutrient prediction, sensor calibration, and precision agriculture applications. It enables comparative studies with conventional spectroscopy-based sensing techniques such as NIR and FTIR while providing a low-cost, visible-range alternative suitable for real-time field deployment.