fpt.desc

来自「数据挖掘相关算法」· DESC 代码 · 共 64 行

/////////////////////////////
README for FP-tree algorithm
////////////////////////////


A. File Description
===================
fpt.c		-- The program file for mining large itemsets using a FP-tree algorithm
		   with user given support threshold.
Makefile	-- Makefile to compile fpt.c.
configSample	-- A sample config. file storing the input parameter for fpt.c.
dataSample.dat  -- A sample data file.


B. Compilation
==============
Type:	make fpt


C. Execution
============
Type:	fpt <config file name>

You can also type "fpt" for details.


D. Data File Format
===================
Each line of the data file consists of one transaction information.
The first value of the line is the number of items in this transaction.
Following the first value are the IDs of items in the transaction.
The IDs of the items are numeric, and is ranged from zero to (M - 1)
where M is the number of items in the database.


E. Config. File Format
======================
Content of config. file:
Line 1: Upper limit of large itemsets size to be mined
        If this value is not larger than zero or
        is greater than the greatest transaction size in the DB,
        then it will be set to the greatest transaction size.
Line 2: Support threshold (normalized to the range of (0, 1])
	The value should be set large enough such that
	any large itemset should appear in at least one transaction.
Line 3: No. of items in the DB
Line 4: No. of transactions in the DB
Line 5: File name of the DB
Line 6: Result file name to store the large itemsets found


F. Running the sample
=====================
Type:	fpt configSample


G. Reference
============
Title:		Mining Frequent Patterns without Candidate Generation
Conference:	SIGMOD 2000
Authors:	Jiawei Han, Jian Pei, and Yiwen Yin