fFilterAssessment

Assessment of filtered observation variants by relating pearson correlation metric (COR), sample number (SN) and optional mean absolute error (MAE=TRUE)

Source

https://github.com/EMRAgit/fPHASE/blob/master/function/fFilterAssessment.R

Parameters

• IN.DIR -> directory of files containing accuracy metrics (result from function fDoyCrit)
• PHASE -> DWD phase ID
• PLANT -> DWD crop type ID
• YEARS -> years of observation
• F.STD -> standard deviation factor
• MAE=TRUE -> mean absolute error is used to assess filtered observation variants

Output

• [OPR-ALL]_[PLANT]-[PHASE].csv -> plant- and phase-specific accuracy metric variants

  • MAE -> Mean Absolute Error
  • MSE -> Mean Squared Error
  • COR -> Pearson correlation coefficient
  • RMSE -> Root Mean Squared Error
  • PLANT -> DWD crop type ID
  • PHASE -> DWD phase ID
  • SN -> Observation sample number
  • STD -> F.STD value
  • YEAR -> year of observation
  • OPT -> result of CORSN (MAE=FALSE) or CORSN/MAE (MAE=TRUE)

• [OPT-MAX]_[PLANT]-[PHASE].csv -> optimal plant-, phase- and year-specific accuracy metric variants with OPT=max

  • MAE -> Mean Absolute Error
  • MSE -> Mean Squared Error
  • COR -> Pearson correlation coefficient
  • RMSE -> Root Mean Squared Error
  • PLANT -> DWD crop type ID
  • PHASE -> DWD phase ID
  • SN -> Observation sample number
  • STD -> F.STD value
  • YEAR -> year of observation
  • OPT -> maximum value of CORSN (MAE=FALSE) or CORSN/MAE (MAE=TRUE)