Every Data Mining Object can be constructed in the following way:

obj = ClassName(constructor_args)
            

Where ClassName is the object's class name and constructor_args is a comma-separated list of constructor arguments. For example, the ClassificationSettings object can be created in the following way:

fs = ClassificationFunctionSettings()
            

The standard way of accessing object properties is with the period, i.e.:

fs.algorithmSettings

The majority of MetaModel objects can be created by using 0-parameter constructors. However, some objects can have constructing parameters. The parameter list is one-to-one mapped to arguments from the Factory.create method for a given MetaModel object. For example, ClassificationTestTask can be created in two ways, because ClassificationTestTaskFactory have two create methods as specified in the javadoc part of the AdvancedMiner Documentation:

create()
create(testDataName, modelName, testResultName)
            

Thus, the object can be created in two ways:

# the first method
task = ClassificationTestTask()
task.testDataName = 'test_data_pd'
task.modelName = 'my_model'
task.testResultName = 'results'

# the second method (the end result is the same):
task = ClassificationTestTask( 'test_data_pd', 'my_model', 'results' )

print "Task created"
    

Task created
    

The save command saves a local object to the Metadata Repository. The saved object must have a unique name (chosen by the user) and an identifier (chosen automatically). If the name is the same as the name of an object already stored in the repository, the old object may be overwritten or an exception will be signaled. The replace parameter controls this behavior. Note that only some objects can be saved to the repository. The saved object must be an instance of PersistentObject.

Syntax:

save(name, object[, replace])

objectExists = 0
physicalData = PhysicalData('someTable')
try:
   save('physicalData', physicalData, 0)
except:
   objectExists = 1

save('physicalData', physicalData)
print 'object saved'
    

object saved
    

The load command loads an object from the Metadata Repository. All changes made to that object are visible only locally. The save function has to be used to store the changed object in the repository.

Syntax:

load(name)

physicalData = load('physicalData')
print 'object loaded'
    

object loaded
    

The rename command changes the name of an object stored in the Metadata Repository.

Syntax:

rename(old_name, new_name)

If the user specifies mralias, it should be the same for both object names. If an object with new_name already exists, an exception will be signalled.

newName = 'model'

rename('modell', newName)
                

The execute command executes tasks stored in the Metadata Repository. This command can work in the blocking (other script commands will wait until the task has been finished) or non-blocking mode. If more then one task is specified, they can be executed synchronously or asynchronously. If any exception occurs during task execution, it will always be displayed in the log window, and, in the case of blocking mode, it will also be visible in the output window. Moreover, if blocking mode is used, an exception will terminate the script.

Syntax:

execute(task, wait = 0 | 1)

where:

task =

name | { sync | async : [ task1, task2, ... ] }

execute('t1')
execute( { sync : ['t2', 't3', { async : [ 't4', 't5' ] } ] }, wait = 0)
                

The delete command deletes an object from the Metadata Repository. All local references to this object remain valid.

Syntax:

delete(name)

delete('model')

The exists function checks if an object with the specified name is already available in the Metadata Repository. It returns a non zero value if it is. If such object does not exists, the value 0 is returned.

Syntax:

exists(name)

if not exists('model'):
   execute('build_task')
                

The terminate command terminates a task or transformation that is running on the server. This function does not wait for the task to terminate and does not guarantee immediate ending of the task. If a task is already terminated or was never run no exception will be thrown.

Syntax:

terminate(name)

terminate('task')
                

The saveEnv function saves a part of the script environment to the Metadata Repository. The following types of object can be saved: float, int, complex, string, dict, list, tuple.

Syntax:

saveEnv(name[, [s1, s2, ...]], isGlobal, replace)

logicalData = 'ld'
physicalData = 'pd'
saveEnv('env', ['logicalData', 'physicalData'])
                

The loadEnv command loads an object with the script variables from the Metadata Repository and sets the loaded variables variables in the current environment.

Syntax:

loadEnv(name)

loadEnv('env')
print physicalData
                

The mrAlias command sets a new Matadata Repository alias. The alias is a binding between a symbolic name and the Metadata Repository. The alias is used by all functions operating on MR objects. The default alias is arbitrally set by the platform. To explicitly set the default mrAlias, use the mrRegistry command.

Syntax:

mrAlias(mrAliasName, server, repository, user, password)

mrAlias('alias_name', '127.0.0.1:2531', '127.0.0.1:2531', 'user', 'password')
                

The log command sends text with the level INFO to the log window.

Syntax:

log(text)

log('Some information')
                

There is an API for more advanced work with Metadata Repository connections. Please refer to RepoReg part of javadoc documentation for details. The function below provides access to a Registry instance.

Syntax:

mrRegistry()

mrRegistry().defaultRepository = 'repository_alias'
                

The projectPath function returns the absolute path to the current project in the repository.

Syntax:

projectPath()

#print the path to the current project
print projectPath()