About

This project is about reaserch and implemention on strategies to find statstical outliers.

List of function implmented

• quantile_
• rank_
• median_
• mad_
• double_mad_
• lm_ // Implementation of R's lm function
• find_outliers_double_mad_based_
• train_n_test_type_1 // Find outlier based on mad
• quantile_iqr_statistics_
• train_n_test_type_2 // Find outlier based on IQR
• train_n_test_type_3 // Find outlier based on linear reggression w/o removing influentials
• hatvalues_
• find_leverage_
• find_influence_cooks_
• find_influence_dffits_
• train_n_test_type_4 // Find outlier based on linear reggression removing influentials
• control_charts_statistics_
• train_n_test_type_5 // Find outlier based on control charts
• quantile_regression_statistics_
• train_n_test_type_6 // Find outlier based on quantile regression based
• train_n_test_type_6_1 # // Find outlier based on quantile regression with cutoff decided by .9 and .1 quantile regression line
• gesd_statistics_
• train_n_test_type_7 // Find outlier based on quantile regression based on gesd statistics
• cusum_statistics_ // link
  • changepoint_analysis_
  • train_n_test_type_8 // Find outlier based on cusum statistics

Usage

source ("driver.R")

Visualization

An example vizuatizaion produced by the script coparing 4 ourlier detection strategies.