Inversion of plant traits using ML models

Usage

hybrid_inversion(
  LUT = NULL,
  input = NULL,
  split = 0.8,
  setseed = NULL,
  method = NULL,
  collinearity = NULL,
  pattern = NULL,
  trans = NULL,
  Field.data = NULL,
  acron = NULL
)

Arguments

LUT: Dataset with inputs and Bands
input: variable to estimate
split: ratio between 0 and 1 for splitting the dataset in training and testing
setseed: set random number
method: Machine learning approach for estimating each plant traits, the options are 'SVM', 'RF' and 'LDA'
collinearity: collinearity-and-stepwise-vif-selection or CARS method implemmented, options='VIF' and 'CARS'
pattern: Please indicate the number of bands with same pattern'B'.
trans: Please indicate is want a logarithm transformation to y variable . Default is T
Field.data: dataframe with observations
acron: acronynm for the observation measure: e.g., Cab_obsrv, where acron='_observ' and Cab has same name as input

Value

A list: model (the fitted ML model), Stats (accuracy statistics on the held-out test split), and Plot (a ggplot scatter of predicted vs. measured/simulated values for input). If Field.data is supplied, the returned Plot/Stats are computed against the field observations instead of the LUT's own test split.