A class that implements the Apriori algorithm for computing itemsets in binary matrix. This class computes closed itemsets. It also computes the lattice connecting the closed itemsets. ! More...

#include <apriori.h>

Inheritance diagram for Apriori:

List of all members.

Public Member Functions
	Apriori (string filename)
	Constructor that reads a binary matrix from a file. Create an instance of Apriori from a file. The file must be in the form of a tab delimited matrix with the first row containing the column names. Each row must begin with the row name. The first token of the first row must exist, but can be any value.
	Apriori (vector< vector< unsigned int > > &matrix, map< unsigned int, string > &majorNames, map< unsigned int, string > &minorNames)
virtual	~Apriori ()
	Destructor.
virtual bool	checkIfClosed (Itemset &itemset) const
virtual void	computeItemsetFromRows (vector< unsigned int > &rowIndices, Itemset &itemset)
	Compute an itemset from a set of rows.
virtual void	computeItemsetFromRows (Itemset &itemset)
	Compute the columns in the itemset given the rows already stored in the itemset.
virtual void	computeItemsets (const set< unsigned int > rowsToAvoidInSeed=NULL, const set< unsigned int > rowsToAvoidCompletely=NULL)
virtual void	computeItemsetsLowRAM (const set< unsigned int > rowsToAvoidInSeed=NULL, const set< unsigned int > rowsToAvoidCompletely=NULL)
	This method is identical to Apriori::computeItemsets() except that it uses less memory by not storing column names with each computed itemset.
virtual AprioriLatticeEdgeType	computeLatticeEdgeType (const Itemset &itm1, const Itemset &itm2) const
virtual void	computeLattice ()
virtual void	computeRandomDistribution (ostream &ostrm, unsigned int numTries, map< ItemsetRandomRowDistKeyType, MyHistogram > &rowDists, map< unsigned int, MyHistogram > &columnDists, MyHistogram &sizeDist)
bool	containsColumn (Itemset &itemset, unsigned int index) const
bool	containsRow (Itemset &itemset, unsigned int index) const
virtual void	getItemsets (vector< Itemset > &itemsets, bool deleteItemsets=false)
virtual void	getLattice (ItemsetLattice &lattice, bool deleteLattice=false)
virtual void	getClosedLattice (ItemsetLattice &closedLattice, bool deleteLattice=false)
	Returns the transitive closure of the lattice connecting the closed itemsets in the data.
virtual unsigned int	getNumItemsets () const
	Return the number of computed itemsets.
virtual unsigned int	getNumLatticeEdges () const
	Return the number of edges in the lattice.
virtual void	printItemsets (ostream &ostr) const
	Print itemsets to the output stream in "itemset" format.
virtual void	printItemsets (string outputFile) const
virtual void	printItemsetsGraph (ostream &ostr) const
	Print the bipartite graph induced by each itemset to the output stream.
virtual void	printItemsetsGraph (string outputFile) const
virtual void	printItemsetStatistics (ostream &ostr) const
	For each k, print the number of itemsets with k rows and with k columns.
void	readFile (string filename)
	Reads a binary matrix into the invocant from a file.
virtual void	setItemsets (const vector< Itemset > &isets)
	Store itemsets (computed by another method) internally.
virtual unsigned int	setMinimums (unsigned int rows, unsigned int columns)
Protected Member Functions
virtual void	_computeAllRows (vector< unsigned int > &allRows) const
virtual unsigned int	_computeRandomRows (const vector< unsigned int > &allRows, vector< unsigned int > &randomRows) const
virtual bool	_areRowsInSameDataset (unsigned int i, unsigned int j) const
Protected Attributes
unsigned int	minrows
unsigned int	mincols
bool	transposed
vector< vector< unsigned int > >	data
map< unsigned int, string >	columnNames
map< unsigned int, string >	rowNames
unsigned int	_numDatasets
vector< unsigned int >	_rowIndexToDatasetIndex
vector< Itemset >	itemsets
bool	_itemsetsComputed
ItemsetLattice	_closedLattice
ItemsetLattice	_reducedLattice
bool	_latticeComputed

Detailed Description

A class that implements the Apriori algorithm for computing itemsets in binary matrix. This class computes closed itemsets. It also computes the lattice connecting the closed itemsets. !

Constructor & Destructor Documentation

Apriori::Apriori ( string filename )

Constructor that reads a binary matrix from a file. Create an instance of Apriori from a file. The file must be in the form of a tab delimited matrix with the first row containing the column names. Each row must begin with the row name. The first token of the first row must exist, but can be any value.

Member Function Documentation

bool Apriori::checkIfClosed ( Itemset & itemset ) const [virtual]

Return true if and only if the itemset is closed.

void Apriori::computeItemsets	(	const set< unsigned int > *	rowsToAvoidInSeed = `NULL`,
		const set< unsigned int > *	rowsToAvoidCompletely = `NULL`
	)		`[virtual]`

Compute closed itemsets in the data using the Apriori algorithm.

Parameters:

[in]	rowsToAvoidInSeed,a	pointer to a set of rows to avoid when computing single-row itemsets. The method starts by computing itemsets with single rows. If a row is a member of this set, the method will not use that row to compute a single-row itemset. The row may be used later in computing itemsets with more than one row.
[in]	rowsToAvoidCompletely,a	pointer to a set of rows to avoid when computing any itemset, including single-row itemsets.

Note:: Use the Apriori::getItemsets() method to retrieve the itemsets.

Reimplemented in AprioriWithComplement.

void Apriori::computeItemsetsLowRAM	(	const set< unsigned int > *	rowsToAvoidInSeed = `NULL`,
		const set< unsigned int > *	rowsToAvoidCompletely = `NULL`
	)		`[virtual]`

This method is identical to Apriori::computeItemsets() except that it uses less memory by not storing column names with each computed itemset.

Note:: You must invoke Apriori::finalizeItemsets() before processing the columns of the computed itemsets.

void Apriori::computeLattice ( ) [virtual]

Compute the lattice induced by subset relationships between the itemsets.

Note:: The method invokes Apriori::computeItemsets() if you have not already invoked it.

AprioriLatticeEdgeType Apriori::computeLatticeEdgeType	(	const Itemset &	itm1,
		const Itemset &	itm2
	)		const `[virtual]`

Compute the type of the edge between two itemsets.

Parameters:

[in]	itm1,an	instance of Itemset.
[in]	itm2,an	instance of Itemset.

Returns:: LATTICE_NO_EDGE if there is no relationship between the two itemsets and LATTICE_EDGE if itm1's rows contain itm2's rows.

Reimplemented in AprioriWithComplement.

void Apriori::computeRandomDistribution	(	ostream &	ostrm,
		unsigned int	numTries,
		map< ItemsetRandomRowDistKeyType, MyHistogram > &	rowDists,
		map< unsigned int, MyHistogram > &	columnDists,
		MyHistogram &	sizeDist
	)		`[virtual]`

Compute a distribution of itemset sizes by picking subsets of rows uniformly at random.

bool Apriori::containsColumn	(	Itemset &	itemset,
		unsigned int	index
	)		const

Return true if and only if the column index should be part of itemset.

The method checks if the column index has a 1 in all the rows currently in itemset.

bool Apriori::containsRow	(	Itemset &	itemset,
		unsigned int	index
	)		const

Return true if and only if the row index should be part of itemset.

The method checks if the row index has a 1 in all the columns currently in itemset.

void Apriori::getClosedLattice	(	ItemsetLattice &	closedLattice,
		bool	deleteLattice = `false`
	)		`[virtual]`

Returns the transitive closure of the lattice connecting the closed itemsets in the data.

Parameters:

[in] deleteLattice,a boolean; if true, delete the closed lattice stored internally.

Note:: The method invokes Apriori::computeLattice() if you have not already invoked it.

void Apriori::getItemsets	(	vector< Itemset > &	itemsets,
		bool	deleteItemsets = `false`
	)		`[virtual]`

Returns a vector of closed itemsets in the data.

Parameters:

[out]	itemsets,a	vector of Itemset to return the closed itemsets in.
[in]	deleteItemsets,a	boolean; if true, delete the itemsets stored internally

Note:: The method invokes Apriori::computeItemsets() if you have not already invoked it.

void Apriori::getLattice	(	ItemsetLattice &	lattice,
		bool	deleteLattice = `false`
	)		`[virtual]`

Returns the lattice connecting the closed itemsets in the data.

Parameters:

[out]	lattice,an	instance of ItemsetLattice to return the lattice in.
[in]	deleteLattice,a	boolean; if true, delete the lattice stored internally.

Note:: The method invokes Apriori::computeLattice() if you have not already invoked it.

void Apriori::printItemsets ( string outputFile ) const [virtual]

Print itemsets to the output file in "itemset" format.

Each itemset appears on one line containing the name of the itemset, the names of the rows in the itemset, and the names of the columns in the itemset, all separated by tabs. The name of the itemset is the string "itemset_<index>_<number of rows>_<number of columns>".

void Apriori::printItemsetsGraph ( string outputFile ) const [virtual]

Print the bipartite graph induced by each itemset to the output file.

Each line of the file contains the itemset name, a row name, and a column name, all separated by tabs.

void Apriori::setItemsets ( const vector< Itemset > & isets ) [virtual]

Store itemsets (computed by another method) internally.

Parameters:

[in] itemsets,a vector of Itemsets to be stored.

Use this method when you have a set of itemsets that you would like to further process using methods in the Apriori class, e.g., Apriori::computeLattice(). A typical use of this method is to compute itemsets using this class, using an external method to prune the itemsets, and computing the lattice involving the pruned itemsets.

Warning:: Using this method will discard any itemsets previously stored in the class.

The documentation for this class was generated from the following files:

/home/poirel/src/c++/biorithm/libapriori/apriori.h
/home/poirel/src/c++/biorithm/libapriori/apriori.C

Public Member Functions

Protected Member Functions

Protected Attributes

Detailed Description

Constructor & Destructor Documentation

Member Function Documentation