blatoo / r_gbod

r_gbod

This is an grid based anomaly detection algorithm, which suitable for large dataset.

Reference Paper: Grid Based Outlier Detection in Large Data Sets for Combine Harvester Author: Ying Gu, Conference: IEEE-INDIN 2017(IEEE 15th International Conference on Industrial Informatics)

Installation

The installation need the devtools package: https://github.com/hadley/devtools

library("devtools")
install_github("blatoo/r_gbod")

Quickstart

Details in file "gbod_usage.html" Input data is a dataframe without missing values

library(dplyr)
library(gbod)
df = na.omit(read.csv(...)) # input dataframe

# use the basic method
res = gbod(df, n_partition=10, outlier_percent=1)
    
# use the scored method
res = gbod_score(df, partition_range = c(5,8), outlier_percent=1)