Overview

{outqrf} is an R package used for outlier detection. Each numeric variable is regressed onto all other variables using a quantile random forest (Meinshausen 2006). We use {ranger} package (Wright and Ziegler 2017) to perform the fitting and prediction of quantile regression forests. Next, we will compute the rank of the observed values in the predicted results’ quantiles. If the rank of the observed value exceeds the threshold, the observed value is considered an outlier.

Since the same predicted value might be distributed across multiple quantiles in the predicted quantile results, this affects our location finding for the observed value. Therefore, we also used a method similar to the outForest package to compare the observed value with the 50% quantile value again to determine the final quantile result.

Installation

# Development version
devtools::install_github("flystar233/outqrf")

Usage

library(outqrf)
#Generate data with outliers in numeric columns
irisWithOutliers <- generateOutliers(iris, p = 0.05,seed =2024)
# Find outliers by quantile random forest regressions
out <- outqrf(irisWithOutliers,quantiles_type=400)
#> 
#> Outlier identification by quantiles random forests
#> 
#>   Variables to check:        Sepal.Length, Sepal.Width, Petal.Length, Petal.Width
#>   Variables used to check:   Sepal.Length, Sepal.Width, Petal.Length, Petal.Width, Species
#> 
#>   Checking: Sepal.Length  Sepal.Width  Petal.Length  Petal.Width
out$outliers
#>    row          col observed predicted   rank
#> 1   32 Sepal.Length     14.9      5.40 0.9975
#> 2   35 Sepal.Length     -1.8      4.60 0.0025
#> 3   84 Sepal.Length     11.5      6.80 0.9975
#> 4  129 Sepal.Length     -5.6      6.30 0.0025
#> 5   49  Sepal.Width     10.8      3.85 0.9975
#> 6  131  Sepal.Width     -2.1      2.70 0.0025
#> 7  137  Sepal.Width     11.5      3.20 0.9975
#> 8   36 Petal.Length     12.8      1.60 0.9975
#> 9   73 Petal.Length    -17.2      4.40 0.0025
#> 10 107 Petal.Length     13.7      5.60 0.9975
#> 11 123 Petal.Length     -9.0      5.20 0.0025
#> 12 140 Petal.Length     13.5      5.80 0.9975
#> 13  10  Petal.Width    -11.8      0.20 0.0025
#> 14  14  Petal.Width     -6.3      0.20 0.0025
#> 15  34  Petal.Width      7.6      0.40 0.9975
#> 16  66  Petal.Width      7.0      2.00 0.9950
#> 17 113  Petal.Width     -6.1      1.80 0.0025

Evaluation on diamonds (Small Dataset)

library(outqrf)
irisWithOutliers <- generateOutliers(iris, p = 0.05,seed =2024)
qrf <- outqrf(irisWithOutliers,quantiles_type=400)
#> 
#> Outlier identification by quantiles random forests
#> 
#>   Variables to check:        Sepal.Length, Sepal.Width, Petal.Length, Petal.Width
#>   Variables used to check:   Sepal.Length, Sepal.Width, Petal.Length, Petal.Width, Species
#> 
#>   Checking: Sepal.Length  Sepal.Width  Petal.Length  Petal.Width

evaluateOutliers(iris,irisWithOutliers,qrf$outliers)
#>     Actual  Predicted      Cover   Coverage Efficiency 
#>      32.00      17.00      17.00       0.53       1.00
plot(qrf)
#> 
#> Outlier identification by quantiles random forests
#> 
#>   Variables to check:        Sepal.Length, Sepal.Width, Petal.Length, Petal.Width
#>   Variables used to check:   Sepal.Length, Sepal.Width, Petal.Length, Petal.Width, Species
#> 
#>   Checking: Sepal.Length  Sepal.Width  Petal.Length  Petal.Width

Evaluation on diamonds (Big Dataset)

library(outqrf)
library(ggplot2)
library(dplyr)
data <- diamonds|>select(price,carat,cut,color,clarity)
data2 <- outqrf::generateOutliers(data, p = 0.001,seed =2024)
# 108
qrf <- outqrf(data2,num.threads=8,quantiles_type=400)
#> 
#> Outlier identification by quantiles random forests
#> 
#>   Variables to check:        price, carat
#>   Variables used to check:   price, carat, cut, color, clarity
#> 
#>   Checking: price  carat
#The process can be slow because it needs to predict the value at 400|1000 quantiles for each observation. 
evaluateOutliers(data,data2,qrf$outliers)
#>     Actual  Predicted      Cover   Coverage Efficiency 
#>     108.00     369.00     103.00       0.95       0.28

References

Meinshausen, Nicolai. 2006. “Quantile Regression Forests.” J. Mach. Learn. Res. 7 (17): 983–99. https://dl.acm.org/doi/10.5555/1248547.1248582.
Wright, Marvin, and Andreas Ziegler. 2017. “Ranger: A Fast Implementation of Random Forests for High Dimensional Data in c++ and r.” Journal of Statistical Software, Articles 77 (1): 1–17. https://doi.org/10.18637/jss.v077.i01.