data-physico-chemical-mecanical

data-physico-chemical-mecanical.xlsx: This dataset consists of 130 soil samples collected from five different provinces, with spectral, physical, chemical, and mechanical properties measured. The samples were analyzed in the laboratory, where physical, chemical, and mechanical soil properties were determined. Physical, chemical, and spectral features were used as model inputs, and mechanical properties were estimated using the model. These data can be used for analyzing the effects of various soil properties, developing predictive models for soil mechanical behavior, and supporting research in soil management and agriculture.. RAW-Spectra-400-2450 nm.rar: This file contains the spectral data preprocessing applied using Parles software. Five different preprocessing methods were applied to the spectral data, which were then used as model inputs. By combining these spectral features with soil properties, a total of 11 transfer functions were generated for estimating soil compaction and swelling indices. Mean Random-RF-MR-RAW-COMPACTION: This file contains the analysis of the 11 transfer functions generated for predicting soil compaction and swelling indices. The analyses were performed using Random Forest and Multiple Linear Regression methods. Error statistics for the models were calculated to evaluate their predictive performance.data-physico-chemical-mecanical.xlsx：本数据集涵盖采自5个不同省份的130份土壤样本，测定了其光谱、物理、化学及力学特性。所有样本均于实验室完成分析，完成土壤物理、化学与力学特性的测定流程。本研究以光谱、物理及化学特征作为模型输入，依托模型实现土壤力学特性的预测。该数据集可用于剖析各类土壤特性的影响效应、构建土壤力学行为预测模型，同时可为土壤管理与农业相关研究提供数据支撑。 RAW-Spectra-400-2450 nm.rar：该压缩包包含经Parles软件预处理的光谱数据。研究人员对原始光谱数据采用了5种不同的预处理方案，将处理得到的光谱特征作为模型输入。将此类光谱特征与土壤特性相结合，共生成11种用于估算土壤压实与膨胀指数的传递函数。 Mean Random-RF-MR-RAW-COMPACTION：该文件包含针对上述11种土壤压实与膨胀指数预测传递函数的分析结果。分析采用随机森林（Random Forest）与多元线性回归（Multiple Linear Regression）两种方法开展，并计算各模型的误差统计量以评估其预测性能。