A greedy algorithm which detects outliers in categorical data based on entropy measure. This is a C++20 implementation of

He Z., Deng S., Xu X., Huang J.Z. (2006) A Fast Greedy Algorithm for Outlier Mining. In: Ng WK., Kitsuregawa M., Li J., Chang K. (eds) Advances in Knowledge Discovery and Data Mining. PAKDD 2006. Lecture Notes in Computer Science, vol 3918. Springer, Berlin, Heidelberg.

A C++20 supported compiler is required. Build the tool with

$ make

./greedy_outlier_entropy -k <value> -d <data-file-path> [-o <output-file-path>]

Arguments:

• -k: Positive integer number of outliers.
• -d: Path to input data file in matrix form. Each row denotes a data point and each column stands for a categorical feature dimension. Data are seperated by a comma. Do not use any whitespaces between feature values.
• -o: (Optional) Path to output file with non-outlier data.

Three data points, each with four feature dimensions.

0,0,1,0
1,0,2,1
2,1,0,2

We can also use arbitrary strings as feature values:

James,a,g,k
Richard,b,d,p
David,b,e,o

A bash script is provided to run the experiment on Lymphography data from the UCI Machine Learning Repository. Simply run

make experiments

• Lymphography: This lymphography domain was obtained from the University Medical Centre, Institute of Oncology, Ljubljana, Yugoslavia. Thanks go to M. Zwitter and M. Soklic for providing the data.

• ProgramOptions.hxx for argument parsing (MIT licensed)

This software is released under the MIT license.