mlbackend_php\processor Class Reference

PHP predictions processor. More...

Inheritance diagram for mlbackend_php\processor:

Public Member Functions
	classify ($uniqueid, stored_file $dataset, $outputdir)
	Classifies the provided dataset samples.
	clear_model ($uniqueid, $modelversionoutputdir)
	Delete the stored models.
	delete_output_dir ($modeloutputdir, $uniqueid)
	Delete the output directory.
	estimate ($uniqueid, stored_file $dataset, $outputdir)
	Estimates linear values for the provided dataset samples.
	evaluate_classification ($uniqueid, $maxdeviation, $niterations, stored_file $dataset, $outputdir, $trainedmodeldir)
	Evaluates this processor classification model using the provided supervised learning dataset.
	evaluate_regression ($uniqueid, $maxdeviation, $niterations, stored_file $dataset, $outputdir, $trainedmodeldir)
	Evaluates this processor regression model using the provided supervised learning dataset.
	export (string $uniqueid, string $modeldir)
	Exports the machine learning model.
	import (string $uniqueid, string $modeldir, string $importdir)
	Imports the provided machine learning model.
	is_ready ()
	Checks if the processor is ready to use.
	train_classification ($uniqueid, stored_file $dataset, $outputdir)
	Train this processor classification model using the provided supervised learning dataset.
	train_regression ($uniqueid, stored_file $dataset, $outputdir)
	Train this processor regression model using the provided supervised learning dataset.

Public Attributes
const	BATCH_SIZE = 5000
	Size of training / prediction batches.
const	MODEL_FILENAME = 'model.ser'
	File name of the serialised model.
const	TRAIN_ITERATIONS = 500
	Number of train iterations.

Protected Member Functions
	extract_metadata ($fh)
	Extracts metadata from the dataset file.
	get_evaluation_result_object (\stored_file $dataset, $scores, $maxdeviation)
	Returns the results objects from all evaluations.
	get_model_filepath (string $modeldir)
	Returns the path to the serialised model file in the provided directory.
	instantiate_algorithm ()
	Instantiates the ML algorithm.
	load_classifier ($outputdir)
	Loads the pre-trained classifier.

Protected Attributes
bool	$limitedsize = false

Detailed Description

PHP predictions processor.

Copyright

2016 David Monllao

License

http://www.gnu.org/copyleft/gpl.html GNU GPL v3 or later

Member Function Documentation

classify()

mlbackend_php\processor::classify	(		$uniqueid,
		stored_file	$dataset,
			$outputdir )

Classifies the provided dataset samples.

Parameters

string	$uniqueid
stored_file	$dataset
string	$outputdir

Return values

stdClass

Implements core_analytics\classifier.

clear_model()

mlbackend_php\processor::clear_model	(		$uniqueid,
			$modelversionoutputdir )

Delete the stored models.

Parameters

string	$uniqueid
string	$modelversionoutputdir

Return values

null

Implements core_analytics\predictor.

delete_output_dir()

mlbackend_php\processor::delete_output_dir	(		$modeloutputdir,
			$uniqueid )

Delete the output directory.

Parameters

string	$modeloutputdir
string	$uniqueid

Return values

null

Implements core_analytics\predictor.

estimate()

mlbackend_php\processor::estimate	(		$uniqueid,
		stored_file	$dataset,
			$outputdir )

Estimates linear values for the provided dataset samples.

Exceptions

new	coding_exception

Parameters

string	$uniqueid
stored_file	$dataset
mixed	$outputdir

Return values

void

Implements core_analytics\regressor.

evaluate_classification()

mlbackend_php\processor::evaluate_classification	(		$uniqueid,
			$maxdeviation,
			$niterations,
		stored_file	$dataset,
			$outputdir,
			$trainedmodeldir )

Evaluates this processor classification model using the provided supervised learning dataset.

During evaluation we need to shuffle the evaluation dataset samples to detect deviated results, if the dataset is massive we can not load everything into memory. We know that 2GB is the minimum memory limit we should have (core_analytics\model\heavy_duty_mode), if we substract the memory that we already consumed and the memory that Phpml algorithms will need we should still have at least 500MB of memory, which should be enough to evaluate a model. In any case this is a robust solution that will work for all sites but it should minimize memory limit problems. Site admins can still set $CFG->mlbackend_php_no_evaluation_limits to true to skip this 500MB limit.

Parameters

string	$uniqueid
float	$maxdeviation
int	$niterations
stored_file	$dataset
string	$outputdir
string	$trainedmodeldir

Return values

stdClass

Implements core_analytics\classifier.

evaluate_regression()

mlbackend_php\processor::evaluate_regression	(		$uniqueid,
			$maxdeviation,
			$niterations,
		stored_file	$dataset,
			$outputdir,
			$trainedmodeldir )

Evaluates this processor regression model using the provided supervised learning dataset.

Exceptions

new	coding_exception

Parameters

string	$uniqueid
float	$maxdeviation
int	$niterations
stored_file	$dataset
string	$outputdir
string	$trainedmodeldir

Return values

stdClass

Implements core_analytics\regressor.

export()

mlbackend_php\processor::export	(	string	$uniqueid,
		string	$modeldir )

Exports the machine learning model.

Exceptions

moodle_exception

Parameters

string	$uniqueid	The model unique id
string	$modeldir	The directory that contains the trained model.

Return values

string

The path to the directory that contains the exported model.

Implements core_analytics\packable.

extract_metadata()

mlbackend_php\processor::extract_metadata ( $fh )

protected

Extracts metadata from the dataset file.

The file poiter should be located at the top of the file.

Parameters

resource

$fh

Return values

array

get_evaluation_result_object()

mlbackend_php\processor::get_evaluation_result_object ( \stored_file $dataset, $scores, $maxdeviation )

protected

Returns the results objects from all evaluations.

Parameters

stored_file	$dataset
array	$scores
float	$maxdeviation

Return values

stdClass

get_model_filepath()

mlbackend_php\processor::get_model_filepath ( string $modeldir )

protected

Returns the path to the serialised model file in the provided directory.

Parameters

string

$modeldir

The model directory

Return values

string

The model file

import()

mlbackend_php\processor::import	(	string	$uniqueid,
		string	$modeldir,
		string	$importdir )

Imports the provided machine learning model.

Parameters

string	$uniqueid	The model unique id
string	$modeldir	The directory that will contain the trained model.
string	$importdir	The directory that contains the files to import.

Return values

bool

Success

Implements core_analytics\packable.

instantiate_algorithm()

mlbackend_php\processor::instantiate_algorithm ( )

protected

Instantiates the ML algorithm.

Return values

Phpml\Classification\Linear\LogisticRegression

is_ready()

mlbackend_php\processor::is_ready

(

)

Checks if the processor is ready to use.

Return values

bool

Implements core_analytics\predictor.

load_classifier()

mlbackend_php\processor::load_classifier ( $outputdir )

protected

Loads the pre-trained classifier.

Exceptions

moodle_exception

Parameters

string

$outputdir

Return values

Phpml\Classification\Linear\LogisticRegression

train_classification()

mlbackend_php\processor::train_classification	(		$uniqueid,
		stored_file	$dataset,
			$outputdir )

Train this processor classification model using the provided supervised learning dataset.

Parameters

string	$uniqueid
stored_file	$dataset
string	$outputdir

Return values

stdClass

Implements core_analytics\classifier.

train_regression()

mlbackend_php\processor::train_regression	(		$uniqueid,
		stored_file	$dataset,
			$outputdir )

Train this processor regression model using the provided supervised learning dataset.

Exceptions

new	coding_exception

Parameters

string	$uniqueid
stored_file	$dataset
string	$outputdir

Return values

stdClass

Implements core_analytics\regressor.

---

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

• lib/mlbackend/php/classes/processor.php