rfpred: Predict O/E scores for new sites

Description Usage Arguments Value Note Author(s)

View source: R/rfpred.r

Description

Predict O/E scores for a new set of sites based on a Random Forest predictive model

Usage

1
2
rfpred(bugcal.pa, grps.final, preds.final, ranfor.mod, prednew, bugnew,
  Pc = 0.5, Cal.OOB = FALSE, bcpred = FALSE)

Arguments

bugcal.pa

data.frame of taxonomy at calibration reference sites

grps.final

data.frame of environmetnal data at calibration reference sites

preds.final

chr string of predictors used in ranfor.mod

ranfor.mod

randomForest model used for prediction

prednew

data.frame containing predictor variables (columns) at all new sites/samples (rows) at which predictions are desired (e.g., 'test' sites)

bugnew

data.frame of taxonomy presence/absence data at all new sites/samples (rows) at which predictions are desired (e.g., 'test' sites)

Pc

numeric value for capture probability cutoff

Cal.OOB

logical value default FALSE, set to TRUE only if you are predicting for the model calibration data and desire out-of-bag predictions

bcpred

logical indicating if BC predictions are added to output

Value

function output is a list containing three elements

Note

Version 4.2 - Added option for out-of-bag predictions on CAL data. 12/30/10

Author(s)

J. Van Sickle, USEPA, [email protected]


SCCWRP/ASCI documentation built on Jan. 12, 2018, 6:33 p.m.