MSApriori implementation of Ferenc Bodon

MSApriori algorithm was originally proposed by Bing Liu, Wynne Hsu and Yiming Ma in "Mining association rules with multiple minimum supports" in 1999 to find frequent itemsets and association rules with multiple minimum supports in a transaction database. Here you can download a fast, trie-based, command-line implementation of the APRIORI algorithm for linux and windows platforms. The code is written in standard C++ in object-oriented manner. It uses STL possibilities, if that does not reduce running speed.

The code can be freely used for research purposes.

Download

Binaries:

  1. to Linux (version 2.2.24, i386, gcc 3.x): msapriori_linux.tar.gz
  2. to Windows (version 2.2.24): msapriori_windows.zip

The zip file contains the program and a configuration file (.apriori_config). Copy the configuration file to the folder where the msapriori program is located.

Source:

  1. Latest version (2.2.24) msapriori.tar.gz (date of issue: 20. Juni 2004.)

To compile in Unix systems do the following:

cd source; make; cd ..

this will result in an msapriori program.

Documentation

Usage

To run msapriori 3 parameters are mandatory:

  1. The file that contain the transactions in market-basket format,
  2. the output file the frequent itemset are sent to,
  3. file that contian relative support thresholds (i.e MIS values).

For example you can type:

./msapriori T40I10D100K.dat ouput.txt mis_of_T40I10D100K.dat

If you want to mine association rules as well, then you can provide a 4th parameter that gives the confidence threshold.
If your input data file is not in market-basket model, look around in the recode_reformat folder.

Useful links

Other FIM implementations: