R/filterData.R

In MNLR/downscaleR.BN: Statistical Downscaling with Bayesian Networks.

#' @export

filterData <- function(data, st1, st2 = NULL, remove.na, evidence.nodes, evidence, silent = TRUE){
  if (is.list(data)){
    Data <- data$Data
    node.names <- data$Metadata$station_id
  } else {
    Data <- data
    node.names <- NULL
  }

  if (is.character(evidence.nodes) && is.null(node.names) ){
    stop("data$Metadata$station_id not found. Please provide evidence nodes as number.")
  }

  N0 <- nrow(Data)

  if (is.character(st1)){ st1<- which(node.names == st1) }
  if (!is.null(st2) && is.character(st2)){ st2<- which(node.names == st2) }
  if (!is.null(evidence.nodes)){
    aux.evidence.nodes <- rep(0,length(evidence.nodes))
    for (i in 1:length(evidence.nodes)){
      if (is.character(evidence.nodes[i])){
        aux.evidence.nodes[i] <- which(node.names == evidence.nodes[i])
      } else { aux.evidence.nodes[i] <- evidence.nodes[i]}
    }
    evidence.nodes <- aux.evidence.nodes
  }

  if (remove.na){ Data <- Data[complete.cases(Data), ] }

  if (!is.null(evidence.nodes)){
    for (i in 1:length(evidence.nodes)){
      Data <- Data[ which(Data[ , evidence.nodes[i]] == evidence[i] ) , ]
    }
  }

  if (!is.null(st2)) sts <- c(st1, st2)
  else sts <- st1

  if (!is.null(nrow(Data)))  Data <- Data[ , sts]
  else (Data <- Data[sts])
  if (!silent){
    N <- nrow(Data)
    print(paste0(N, " observations remaining."))
  }
  return(Data)
}