R/creepy.trend.R

In pawelkasperkiewicz/FewTools: Few practical tools from simple totals in summary to advanced generators for deep learning

Creepy.trend <- function(vector, k, parallel = FALSE) {

  if (parallel) {

    Clust <- makeCluster (detectCores(), type = "PSOCK")
    registerDoParallel(Clust)
    clusterCall(Clust, function(){library(RcppArmadillo)})
    table <- data.frame(matrix(nrow = length(vector), ncol = (length(vector) - k + 1)))
    X <- 1:length(vector)
    smoothed.values <- c()

    table <- foreach (i = 1:(length(vector) - k + 1), .combine = data.frame) %dopar%  {

      model <- fastLm(vector[i:(k + i - 1)], X[i:(k + i - 1)])
      begin <- sample(NA, i - 1, replace = TRUE)
      end <- sample(NA, length(vector) - k - length(begin), replace = TRUE)
      table <- c(begin, model$fitted.values, end)
    }


    smoothed.values <- apply(table, 1, mean, na.rm = TRUE)

    smoothed.values <- apply(table, 1, mean, na.rm = TRUE)

    stopCluster(Clust)
    gc(verbose = F)
    return(smoothed.values)
  }


  else {
    table <- data.frame()
    X <- 1:length(vector)
    smoothed.values <- c()

    for (i in 1:(length(vector) - k + 1)) {

      model <- fastLm(vector[i:(k + i - 1)], X[i:(k + i - 1)])

      table[i:(i+k-1), i] <- model$fitted.values
    }

    smoothed.values <- apply(table, 1, mean, na.rm = TRUE)

    return(smoothed.values)
  }
}