do.rf: A random forest based disease forecasting approach
In akeyel/rf1: Random Forest Workflow 1 for DFMIP

Description Usage Arguments Details Value

The do.rf function runs the random forest portion of the code

do.rf(trap.data, dep.var, independent.vars, results.path, do.spatial = 0,
  create.ci.null = 0, label = "", response.type = "continuous",
  exploratory = TRUE, input.seed = 20180830, temporal.field = "year",
  spatial.field = "location", quantile.model = 1,
  display.messages = 1)

`trap.data`	A data frame containing human cases, mosquito infection rate (if applicable), and independent variables for prediction of human cases and/or mosquito infection rates
`dep.var`	The field name containing the dependent variable to analyze
`independent.vars`	The field names of the independent variables to include in the analysis Variables not listed here but present in the data frame will be excluded from the analysis.
`results.path`	The base path in which to place the modeling results. Some models will create sub-folders for model specific results
`do.spatial`	Whether or not to do spatial crossvalidation (can be time-consuming)
`create.ci.null`	Whether or not to generate null model results
`label`	A label for the analysis run
`response.type`	Whether data should be treated as continuous (mosquito rates, number of cases) or binary (0 or 1).
`exploratory`	Whether to identify the best model (exploratory = TRUE), or whether to run the random forest with all input independent variables (exploratory = FALSE)
`input.seed`	A seed for the random number generator to ensure that the results are repeatable
`temporal.field`	The field containing the temporal units
`spatial.field`	The field containing the spatial units
`quantile.model`	Whether (1) or not (0) to use a quantile random forest for the final model output. All other calculations and model fitting use the standard randomForest package.
`display.messages`	Whether or not update messages should be output

Majority of code was writen in wnv_hlpr.R; January - June 2018. Transferred to this file & modified beginning in March 2019. Code adapted to use common inputs with the ArboMAP code of Davis & Wimberly

A list containing:

MODEL	the random forest prediction model
TEMPORAL.ACCURACY	an accuracy assessment using leave one year out cross-validation
SPATIAL.ACCURACY	an accuracy assessment based on leave one location out cross-validation
TEMPORAL.NULL	Null model results based on estimates across time
SPATIAL.NULL	Null model results based on estimates across space
RETAINED.VARS	Variables retained in the final prediction model