data-physico-chemical-mecanical
Description
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.
Files
Steps to reproduce
1- Collect 130 soil samples from the specified five provinces. 2- Measure the physical, chemical, and mechanical properties of the soil samples in the laboratory. 3- Acquire spectral data for each sample in the 400–2450 nm range. 4- Apply five different spectral preprocessing methods using Parles software. Combine the preprocessed spectral features with the physical and chemical soil properties to generate 11 transfer functions. 5- Use Random Forest and Multiple Linear Regression methods to model and predict soil compaction and swelling indices. 6- Calculate error statistics to evaluate the performance of the models.