R/remove_yearly_sin_fit.R

In td-berlin/anomalizer: Implementation of Outlier-Detection-Methods.

#' Remove sin-curve fit over a year
#' @param dates vector of dates of TS
#' @param values vector of values of TS
#' @param max.p.value p.value up to which sin-curve is removed
#' @return vector vector of values
#' @export
remove_yearly_sin_fit <- function(dates, values, max.p.value = 0.1){
  timeline <- data.frame(Date = seq.Date(from = min(dates), to = max(dates), by = "day"))
  df.here <- timeline %>% left_join(
    data.frame(Date = dates, Value = values)) %>% mutate(helper = 1) %>%
    mutate(t = cumsum(helper)) %>% select(-helper)
  models <- list()
  for(i in 1:6){
    df.use <- df.here
    df.use$t <- df.here$t + i * (365.5/12)
    df.use %<>% na.omit
    models[[i]] <- lm(Value ~ sin(2*pi/365.5*t), data = df.use)
  }
  take <- map(models, AIC) %>% which.min
  if(summary(models[[take]])$coefficients[2,4] > max.p.value){
    return(values) } else {
      res_plus_intercept = models[[take]]$residuals + models[[take]]$coefficients[1]
      return(res_plus_intercept)
    }
}