Table 1.
Method for measurement of chemical parameters used in Texas A&M Soil, Water and Forage Testing Laboratory, College Station, TX [27–35].
Fig 1.
A pipeline of the approach used in the paper for prescribing recommendation rules.
Table 2.
Feature importance values of the predictors in the analysis given by the ExtraTreesClassifier.
Table 3.
Synthetic data generation using the MC technique where the mean and covariance matrices are not shared between the classes.
Table 4.
Synthetic data generation using the MC technique where the mean and covariance matrices are not shared between the classes.
Fig 2.
Classification results using baseline model (without any feature engineering techniques).
Fig 3.
Classification results using quantile transformation on the dataset (applying normal quantile distribution on the dataset with the number of quantiles set to 100 and output distribution as uniform).
Fig 4.
Classification results using power transformation on the numerical predictors and later appending the categorical predictors to the dataset.
Fig 5.
Classification results using power transformation on the numerical predictors and clipping the lowest and highest quantiles of data.
Fig 6.
Classification results using gaussian transformation on the dataset after ranking the numerical predictors.
Fig 7.
Decision tree stating the recommended rules based on the output from the Machine Learning system.