com.bayesia.api

Interface Database

public interface Database

This interface must be implemented in order to perform network learning. The learning process needs to know, for each example in the database, the index of the state or interval for each discrete or continuous variable as well as the weight associated to each example. The missing values are returned as -1. This weight must be superior or equals to 0.

The database must provide an instance of NetworkDescriptor. This descriptor allows the APILearning to create a brand new network corresponding to the database or to check the compatibility between the database and an existing network. It must exactly fit the database i.e. the order of the variables must be the same as the one returned when the learning process retrieves the states in each example.

Version:

4.2

Author:

Bayesia S.A.S.

Method Summary

Modifier and Type	Method and Description
short[]	getExampleAt(int row) Return an array of short corresponding to the index of each variable's state found in the database at this row.
double	getExampleValueAt(int row, int variableIndex) Return the numerical value at the given row for the given variable's index.
NetworkDescriptor	getNetworkDescriptor() Return a NetworkDescriptor corresponding to this database.
double	getTotalWeight() Return the sum of all the weights in the database.
double	getWeightAt(int row) Return the weight of the example at the given row.
boolean	hasMissingValues() Tests if the database contains at least one missing value
int	size() Return the size of the database, i.e. the number of examples in the database.

Method Detail

getNetworkDescriptor

NetworkDescriptor getNetworkDescriptor()

Return a NetworkDescriptor corresponding to this database.

Returns:

a network's descriptor

size

int size()

Return the size of the database, i.e. the number of examples in the database.

Returns:

the size of the database

getExampleAt

short[] getExampleAt(int row)

Return an array of short corresponding to the index of each variable's state found in the database at this row. If there is a missing value in a column, the returned value must be -1 at the corresponding index.

Parameters:

row - the index of the row we want the example

Returns:

an array of indices corresponding to the nodes' states or -1 if the value is missing

getExampleValueAt

double getExampleValueAt(int row,
                         int variableIndex)

Return the numerical value at the given row for the given variable's index. This method is called only on continuous variables. If the corresponding value is missing, Double.NaN must be returned.

Parameters:

row - the index of the row in database

variableIndex - the index of a continuous variable

Returns:

the numerical value of the variable in the database at the given row

getWeightAt

double getWeightAt(int row)

Return the weight of the example at the given row. It must return the same value (1 for example) if no weight is associated. The weight must be superior or equals to 0. The sum of all the weights in the database must be superior to 0

Parameters:

row - the index of the row

Returns:

a positive double representing the weight of this example in the database

getTotalWeight

double getTotalWeight()

Return the sum of all the weights in the database. It must be strictly positive.

Returns:

the total weight of the database

hasMissingValues

boolean hasMissingValues()

Tests if the database contains at least one missing value

Returns:

true if the database contains at least one missing value