Qfyilyi

I've written a function using this article (http://www.questionflow.org/2017/12/26/combined-outlier-detection-with-dplyr-and-ruler/) I found online which flags any outliers within my tbl based on either being outside of z-score threshold, or outside of 1.5*IQR. I want to enhance my function by adding an "impute" argument which can be set to mean, median, or mode, (or none, in which case the output is just a tbl of logicals) which will set the values that were flagged to the input argument for that column. How can I go about doing that? My code so far:

# Detecting outliers based on two outlier metrics (3 out of z-score or 1.5 out of IQR)

# Defining the functions to detect outliers

isnt_out_z <- function(x, thres = 3, na.rm = TRUE) {

  abs(x - mean(x, na.rm = na.rm)) <= thres * sd(x, na.rm = na.rm)

}



isnt_out_IQR <- function(x, k = 1.5, na.rm = TRUE) {

  quar <- quantile(x, probs = c(0.25, 0.75), na.rm = na.rm)

  iqr <- diff(quar)

  (quar[1] - k * iqr <= x) & (x <= quar[2] + k * iqr)

}



# Column-based non-outlier rows: row is not an outlier based on some column if it doesn't contain outlier (computed based on target column) on the intersection with that column



## Useable functions

find_Outliers <- function(data, method = 'z', impute = none) {

  if (method == 'z') {

    data %>%

      transmute_if(is.numeric, isnt_out_z)

  } else {

    data %>%

      transmute_if(is.numeric, isnt_out_IQR)

  }

}