1 |
efrain |
1 |
<?php
|
|
|
2 |
// This file is part of Moodle - http://moodle.org/
|
|
|
3 |
//
|
|
|
4 |
// Moodle is free software: you can redistribute it and/or modify
|
|
|
5 |
// it under the terms of the GNU General Public License as published by
|
|
|
6 |
// the Free Software Foundation, either version 3 of the License, or
|
|
|
7 |
// (at your option) any later version.
|
|
|
8 |
//
|
|
|
9 |
// Moodle is distributed in the hope that it will be useful,
|
|
|
10 |
// but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
11 |
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
|
12 |
// GNU General Public License for more details.
|
|
|
13 |
//
|
|
|
14 |
// You should have received a copy of the GNU General Public License
|
|
|
15 |
// along with Moodle. If not, see <http://www.gnu.org/licenses/>.
|
|
|
16 |
|
|
|
17 |
/**
|
|
|
18 |
* Datasets manager.
|
|
|
19 |
*
|
|
|
20 |
* @package core_analytics
|
|
|
21 |
* @copyright 2016 David Monllao {@link http://www.davidmonllao.com}
|
|
|
22 |
* @license http://www.gnu.org/copyleft/gpl.html GNU GPL v3 or later
|
|
|
23 |
*/
|
|
|
24 |
|
|
|
25 |
namespace core_analytics;
|
|
|
26 |
|
|
|
27 |
defined('MOODLE_INTERNAL') || die();
|
|
|
28 |
|
|
|
29 |
/**
|
|
|
30 |
* Datasets manager.
|
|
|
31 |
*
|
|
|
32 |
* @package core_analytics
|
|
|
33 |
* @copyright 2016 David Monllao {@link http://www.davidmonllao.com}
|
|
|
34 |
* @license http://www.gnu.org/copyleft/gpl.html GNU GPL v3 or later
|
|
|
35 |
*/
|
|
|
36 |
class dataset_manager {
|
|
|
37 |
|
|
|
38 |
/**
|
|
|
39 |
* File area for labelled datasets.
|
|
|
40 |
*/
|
|
|
41 |
const LABELLED_FILEAREA = 'labelled';
|
|
|
42 |
|
|
|
43 |
/**
|
|
|
44 |
* File area for unlabelled datasets.
|
|
|
45 |
*/
|
|
|
46 |
const UNLABELLED_FILEAREA = 'unlabelled';
|
|
|
47 |
|
|
|
48 |
/**
|
|
|
49 |
* File area for exported datasets.
|
|
|
50 |
*/
|
|
|
51 |
const EXPORT_FILEAREA = 'export';
|
|
|
52 |
|
|
|
53 |
/**
|
|
|
54 |
* Evaluation file file name.
|
|
|
55 |
*/
|
|
|
56 |
const EVALUATION_FILENAME = 'evaluation.csv';
|
|
|
57 |
|
|
|
58 |
/**
|
|
|
59 |
* The model id.
|
|
|
60 |
*
|
|
|
61 |
* @var int
|
|
|
62 |
*/
|
|
|
63 |
protected $modelid;
|
|
|
64 |
|
|
|
65 |
/**
|
|
|
66 |
* Range processor in use.
|
|
|
67 |
*
|
|
|
68 |
* @var string
|
|
|
69 |
*/
|
|
|
70 |
protected $timesplittingid;
|
|
|
71 |
|
|
|
72 |
/**
|
|
|
73 |
* @var int
|
|
|
74 |
*/
|
|
|
75 |
protected $analysableid;
|
|
|
76 |
|
|
|
77 |
/**
|
|
|
78 |
* Whether this is a dataset for evaluation or not.
|
|
|
79 |
*
|
|
|
80 |
* @var bool
|
|
|
81 |
*/
|
|
|
82 |
protected $evaluation;
|
|
|
83 |
|
|
|
84 |
/**
|
|
|
85 |
* The dataset filearea. Must be one of the self::*_FILEAREA options.
|
|
|
86 |
*
|
|
|
87 |
* @var string
|
|
|
88 |
*/
|
|
|
89 |
protected $filearea;
|
|
|
90 |
|
|
|
91 |
/**
|
|
|
92 |
* Constructor method.
|
|
|
93 |
*
|
|
|
94 |
* @throws \coding_exception
|
|
|
95 |
* @param int $modelid
|
|
|
96 |
* @param int $analysableid
|
|
|
97 |
* @param string $timesplittingid
|
|
|
98 |
* @param string $filearea
|
|
|
99 |
* @param bool $evaluation
|
|
|
100 |
* @return void
|
|
|
101 |
*/
|
|
|
102 |
public function __construct($modelid, $analysableid, $timesplittingid, $filearea, $evaluation = false) {
|
|
|
103 |
|
|
|
104 |
if ($filearea !== self::EXPORT_FILEAREA && $filearea !== self::LABELLED_FILEAREA &&
|
|
|
105 |
$filearea !== self::UNLABELLED_FILEAREA) {
|
|
|
106 |
throw new \coding_exception('Invalid provided filearea');
|
|
|
107 |
}
|
|
|
108 |
|
|
|
109 |
$this->modelid = $modelid;
|
|
|
110 |
$this->analysableid = $analysableid;
|
|
|
111 |
$this->timesplittingid = $timesplittingid;
|
|
|
112 |
$this->filearea = $filearea;
|
|
|
113 |
$this->evaluation = $evaluation;
|
|
|
114 |
}
|
|
|
115 |
|
|
|
116 |
/**
|
|
|
117 |
* Store the dataset in the internal file system.
|
|
|
118 |
*
|
|
|
119 |
* @param array $data
|
|
|
120 |
* @return \stored_file
|
|
|
121 |
*/
|
|
|
122 |
public function store($data) {
|
|
|
123 |
|
|
|
124 |
// Delete previous file if it exists.
|
|
|
125 |
$fs = get_file_storage();
|
|
|
126 |
|
|
|
127 |
$filerecord = [
|
|
|
128 |
'component' => 'analytics',
|
|
|
129 |
'filearea' => $this->filearea,
|
|
|
130 |
'itemid' => $this->modelid,
|
|
|
131 |
'contextid' => \context_system::instance()->id,
|
|
|
132 |
'filepath' => '/analysable/' . $this->analysableid . '/' .
|
|
|
133 |
\core_analytics\analysis::clean_time_splitting_id($this->timesplittingid) . '/',
|
|
|
134 |
'filename' => self::get_filename($this->evaluation)
|
|
|
135 |
];
|
|
|
136 |
|
|
|
137 |
// Delete previous and old (we already checked that previous copies are not recent) evaluation files for this analysable.
|
|
|
138 |
if ($this->evaluation) {
|
|
|
139 |
$select = " = {$filerecord['itemid']} AND filepath = :filepath";
|
|
|
140 |
$fs->delete_area_files_select($filerecord['contextid'], $filerecord['component'], $filerecord['filearea'],
|
|
|
141 |
$select, array('filepath' => $filerecord['filepath']));
|
|
|
142 |
}
|
|
|
143 |
|
|
|
144 |
// Write all this stuff to a tmp file.
|
|
|
145 |
$filepath = make_request_directory() . DIRECTORY_SEPARATOR . $filerecord['filename'];
|
|
|
146 |
$fh = fopen($filepath, 'w+');
|
|
|
147 |
if (!$fh) {
|
|
|
148 |
return false;
|
|
|
149 |
}
|
|
|
150 |
foreach ($data as $line) {
|
|
|
151 |
fputcsv($fh, $line);
|
|
|
152 |
}
|
|
|
153 |
fclose($fh);
|
|
|
154 |
|
|
|
155 |
return $fs->create_file_from_pathname($filerecord, $filepath);
|
|
|
156 |
}
|
|
|
157 |
|
|
|
158 |
/**
|
|
|
159 |
* Returns the previous evaluation file.
|
|
|
160 |
*
|
|
|
161 |
* Important to note that this is per modelid + timesplittingid, when dealing with multiple
|
|
|
162 |
* analysables this is the merged file. Do not confuse with self::get_evaluation_analysable_file
|
|
|
163 |
*
|
|
|
164 |
* @param int $modelid
|
|
|
165 |
* @param string $timesplittingid
|
|
|
166 |
* @return \stored_file
|
|
|
167 |
*/
|
|
|
168 |
public static function get_previous_evaluation_file($modelid, $timesplittingid) {
|
|
|
169 |
$fs = get_file_storage();
|
|
|
170 |
// Evaluation data is always labelled.
|
|
|
171 |
$filepath = '/timesplitting/' . \core_analytics\analysis::clean_time_splitting_id($timesplittingid) . '/';
|
|
|
172 |
return $fs->get_file(\context_system::instance()->id, 'analytics', self::LABELLED_FILEAREA, $modelid,
|
|
|
173 |
$filepath, self::EVALUATION_FILENAME);
|
|
|
174 |
}
|
|
|
175 |
|
|
|
176 |
/**
|
|
|
177 |
* Gets the list of files that couldn't be previously used for training and prediction.
|
|
|
178 |
*
|
|
|
179 |
* @param int $modelid
|
|
|
180 |
* @param bool $includetarget
|
|
|
181 |
* @param string[] $timesplittingids
|
|
|
182 |
* @return null
|
|
|
183 |
*/
|
|
|
184 |
public static function get_pending_files($modelid, $includetarget, $timesplittingids) {
|
|
|
185 |
global $DB;
|
|
|
186 |
|
|
|
187 |
$fs = get_file_storage();
|
|
|
188 |
|
|
|
189 |
if ($includetarget) {
|
|
|
190 |
$filearea = self::LABELLED_FILEAREA;
|
|
|
191 |
$usedfileaction = 'trained';
|
|
|
192 |
} else {
|
|
|
193 |
$filearea = self::UNLABELLED_FILEAREA;
|
|
|
194 |
$usedfileaction = 'predicted';
|
|
|
195 |
}
|
|
|
196 |
|
|
|
197 |
$select = 'modelid = :modelid AND action = :action';
|
|
|
198 |
$params = array('modelid' => $modelid, 'action' => $usedfileaction);
|
|
|
199 |
$usedfileids = $DB->get_fieldset_select('analytics_used_files', 'fileid', $select, $params);
|
|
|
200 |
|
|
|
201 |
// Very likely that we will only have 1 time splitting method here.
|
|
|
202 |
$filesbytimesplitting = array();
|
|
|
203 |
foreach ($timesplittingids as $timesplittingid) {
|
|
|
204 |
|
|
|
205 |
$filepath = '/timesplitting/' . \core_analytics\analysis::clean_time_splitting_id($timesplittingid) . '/';
|
|
|
206 |
$files = $fs->get_directory_files(\context_system::instance()->id, 'analytics', $filearea, $modelid, $filepath);
|
|
|
207 |
foreach ($files as $file) {
|
|
|
208 |
|
|
|
209 |
// Discard evaluation files.
|
|
|
210 |
if ($file->get_filename() === self::EVALUATION_FILENAME) {
|
|
|
211 |
continue;
|
|
|
212 |
}
|
|
|
213 |
|
|
|
214 |
// No dirs.
|
|
|
215 |
if ($file->is_directory()) {
|
|
|
216 |
continue;
|
|
|
217 |
}
|
|
|
218 |
|
|
|
219 |
// Already used for training.
|
|
|
220 |
if (in_array($file->get_id(), $usedfileids)) {
|
|
|
221 |
continue;
|
|
|
222 |
}
|
|
|
223 |
|
|
|
224 |
$filesbytimesplitting[$timesplittingid][] = $file;
|
|
|
225 |
}
|
|
|
226 |
}
|
|
|
227 |
|
|
|
228 |
return $filesbytimesplitting;
|
|
|
229 |
}
|
|
|
230 |
|
|
|
231 |
/**
|
|
|
232 |
* Deletes previous evaluation files of this model.
|
|
|
233 |
*
|
|
|
234 |
* @param int $modelid
|
|
|
235 |
* @param string $timesplittingid
|
|
|
236 |
* @return bool
|
|
|
237 |
*/
|
|
|
238 |
public static function delete_previous_evaluation_file($modelid, $timesplittingid) {
|
|
|
239 |
if ($file = self::get_previous_evaluation_file($modelid, $timesplittingid)) {
|
|
|
240 |
$file->delete();
|
|
|
241 |
return true;
|
|
|
242 |
}
|
|
|
243 |
|
|
|
244 |
return false;
|
|
|
245 |
}
|
|
|
246 |
|
|
|
247 |
/**
|
|
|
248 |
* Returns this (model + analysable + time splitting) file.
|
|
|
249 |
*
|
|
|
250 |
* @param int $modelid
|
|
|
251 |
* @param int $analysableid
|
|
|
252 |
* @param string $timesplittingid
|
|
|
253 |
* @return \stored_file
|
|
|
254 |
*/
|
|
|
255 |
public static function get_evaluation_analysable_file($modelid, $analysableid, $timesplittingid) {
|
|
|
256 |
|
|
|
257 |
// Delete previous file if it exists.
|
|
|
258 |
$fs = get_file_storage();
|
|
|
259 |
|
|
|
260 |
// Always evaluation.csv and labelled as it is an evaluation file.
|
|
|
261 |
$filearea = self::LABELLED_FILEAREA;
|
|
|
262 |
$filename = self::get_filename(true);
|
|
|
263 |
$filepath = '/analysable/' . $analysableid . '/' .
|
|
|
264 |
\core_analytics\analysis::clean_time_splitting_id($timesplittingid) . '/';
|
|
|
265 |
return $fs->get_file(\context_system::instance()->id, 'analytics', $filearea, $modelid, $filepath, $filename);
|
|
|
266 |
}
|
|
|
267 |
|
|
|
268 |
/**
|
|
|
269 |
* Merge multiple files into one.
|
|
|
270 |
*
|
|
|
271 |
* Important! It is the caller responsability to ensure that the datasets are compatible.
|
|
|
272 |
*
|
|
|
273 |
* @param array $files
|
|
|
274 |
* @param int $modelid
|
|
|
275 |
* @param string $timesplittingid
|
|
|
276 |
* @param string $filearea
|
|
|
277 |
* @param bool $evaluation
|
|
|
278 |
* @return \stored_file
|
|
|
279 |
*/
|
|
|
280 |
public static function merge_datasets(array $files, $modelid, $timesplittingid, $filearea, $evaluation = false) {
|
|
|
281 |
|
|
|
282 |
$tmpfilepath = make_request_directory() . DIRECTORY_SEPARATOR . 'tmpfile.csv';
|
|
|
283 |
|
|
|
284 |
// Add headers.
|
|
|
285 |
// We could also do this with a single iteration gathering all files headers and appending them to the beginning of the file
|
|
|
286 |
// once all file contents are merged.
|
|
|
287 |
$varnames = '';
|
|
|
288 |
$analysablesvalues = array();
|
|
|
289 |
foreach ($files as $file) {
|
|
|
290 |
$rh = $file->get_content_file_handle();
|
|
|
291 |
|
|
|
292 |
// Copy the var names as they are, all files should have the same var names.
|
|
|
293 |
$varnames = fgetcsv($rh);
|
|
|
294 |
|
|
|
295 |
$analysablesvalues[] = fgetcsv($rh);
|
|
|
296 |
|
|
|
297 |
// Copy the columns as they are, all files should have the same columns.
|
|
|
298 |
$columns = fgetcsv($rh);
|
|
|
299 |
}
|
|
|
300 |
|
|
|
301 |
// Merge analysable values skipping the ones that are the same in all analysables.
|
|
|
302 |
$values = array();
|
|
|
303 |
foreach ($analysablesvalues as $analysablevalues) {
|
|
|
304 |
foreach ($analysablevalues as $varkey => $value) {
|
|
|
305 |
// Sha1 to make it unique.
|
|
|
306 |
$values[$varkey][sha1($value)] = $value;
|
|
|
307 |
}
|
|
|
308 |
}
|
|
|
309 |
foreach ($values as $varkey => $varvalues) {
|
|
|
310 |
$values[$varkey] = implode('|', $varvalues);
|
|
|
311 |
}
|
|
|
312 |
|
|
|
313 |
// Start writing to the merge file.
|
|
|
314 |
$wh = fopen($tmpfilepath, 'w');
|
|
|
315 |
if (!$wh) {
|
|
|
316 |
throw new \moodle_exception('errorcannotwritedataset', 'analytics', '', $tmpfilepath);
|
|
|
317 |
}
|
|
|
318 |
|
|
|
319 |
fputcsv($wh, $varnames);
|
|
|
320 |
fputcsv($wh, $values);
|
|
|
321 |
fputcsv($wh, $columns);
|
|
|
322 |
|
|
|
323 |
// Iterate through all files and add them to the tmp one. We don't want file contents in memory.
|
|
|
324 |
foreach ($files as $file) {
|
|
|
325 |
$rh = $file->get_content_file_handle();
|
|
|
326 |
|
|
|
327 |
// Skip headers.
|
|
|
328 |
fgets($rh);
|
|
|
329 |
fgets($rh);
|
|
|
330 |
fgets($rh);
|
|
|
331 |
|
|
|
332 |
// Copy all the following lines.
|
|
|
333 |
while ($line = fgets($rh)) {
|
|
|
334 |
fwrite($wh, $line);
|
|
|
335 |
}
|
|
|
336 |
fclose($rh);
|
|
|
337 |
}
|
|
|
338 |
fclose($wh);
|
|
|
339 |
|
|
|
340 |
$filerecord = [
|
|
|
341 |
'component' => 'analytics',
|
|
|
342 |
'filearea' => $filearea,
|
|
|
343 |
'itemid' => $modelid,
|
|
|
344 |
'contextid' => \context_system::instance()->id,
|
|
|
345 |
'filepath' => '/timesplitting/' . \core_analytics\analysis::clean_time_splitting_id($timesplittingid) . '/',
|
|
|
346 |
'filename' => self::get_filename($evaluation)
|
|
|
347 |
];
|
|
|
348 |
|
|
|
349 |
$fs = get_file_storage();
|
|
|
350 |
|
|
|
351 |
return $fs->create_file_from_pathname($filerecord, $tmpfilepath);
|
|
|
352 |
}
|
|
|
353 |
|
|
|
354 |
/**
|
|
|
355 |
* Exports the model training data.
|
|
|
356 |
*
|
|
|
357 |
* @param int $modelid
|
|
|
358 |
* @param string $timesplittingid
|
|
|
359 |
* @return \stored_file|false
|
|
|
360 |
*/
|
|
|
361 |
public static function export_training_data($modelid, $timesplittingid) {
|
|
|
362 |
|
|
|
363 |
$fs = get_file_storage();
|
|
|
364 |
|
|
|
365 |
$contextid = \context_system::instance()->id;
|
|
|
366 |
$filepath = '/timesplitting/' . \core_analytics\analysis::clean_time_splitting_id($timesplittingid) . '/';
|
|
|
367 |
|
|
|
368 |
$files = $fs->get_directory_files($contextid, 'analytics', self::LABELLED_FILEAREA, $modelid,
|
|
|
369 |
$filepath, true, false);
|
|
|
370 |
|
|
|
371 |
// Discard evaluation files.
|
|
|
372 |
foreach ($files as $key => $file) {
|
|
|
373 |
if ($file->get_filename() === self::EVALUATION_FILENAME) {
|
|
|
374 |
unset($files[$key]);
|
|
|
375 |
}
|
|
|
376 |
}
|
|
|
377 |
|
|
|
378 |
if (empty($files)) {
|
|
|
379 |
return false;
|
|
|
380 |
}
|
|
|
381 |
|
|
|
382 |
return self::merge_datasets($files, $modelid, $timesplittingid, self::EXPORT_FILEAREA);
|
|
|
383 |
}
|
|
|
384 |
|
|
|
385 |
/**
|
|
|
386 |
* Returns the dataset file data structured by sampleids using the indicators and target column names.
|
|
|
387 |
*
|
|
|
388 |
* @param \stored_file $dataset
|
|
|
389 |
* @return array
|
|
|
390 |
*/
|
|
|
391 |
public static function get_structured_data(\stored_file $dataset) {
|
|
|
392 |
|
|
|
393 |
if ($dataset->get_filearea() !== 'unlabelled') {
|
|
|
394 |
throw new \coding_exception('Sorry, only support for unlabelled data');
|
|
|
395 |
}
|
|
|
396 |
|
|
|
397 |
$rh = $dataset->get_content_file_handle();
|
|
|
398 |
|
|
|
399 |
// Skip dataset info.
|
|
|
400 |
fgets($rh);
|
|
|
401 |
fgets($rh);
|
|
|
402 |
|
|
|
403 |
$calculations = array();
|
|
|
404 |
|
|
|
405 |
$headers = fgetcsv($rh);
|
|
|
406 |
// Get rid of the sampleid column name.
|
|
|
407 |
array_shift($headers);
|
|
|
408 |
|
|
|
409 |
while ($columns = fgetcsv($rh)) {
|
|
|
410 |
$uniquesampleid = array_shift($columns);
|
|
|
411 |
|
|
|
412 |
// Unfortunately fgetcsv does not respect line's var types.
|
|
|
413 |
$calculations[$uniquesampleid] = array_map(function($value) {
|
|
|
414 |
|
|
|
415 |
if ($value === '') {
|
|
|
416 |
// We really want them as null because converted to float become 0
|
|
|
417 |
// and we need to treat the values separately.
|
|
|
418 |
return null;
|
|
|
419 |
} else if (is_numeric($value)) {
|
|
|
420 |
return floatval($value);
|
|
|
421 |
}
|
|
|
422 |
return $value;
|
|
|
423 |
}, array_combine($headers, $columns));
|
|
|
424 |
}
|
|
|
425 |
|
|
|
426 |
return $calculations;
|
|
|
427 |
}
|
|
|
428 |
|
|
|
429 |
/**
|
|
|
430 |
* Delete all files of a model.
|
|
|
431 |
*
|
|
|
432 |
* @param int $modelid
|
|
|
433 |
* @return bool
|
|
|
434 |
*/
|
|
|
435 |
public static function clear_model_files($modelid) {
|
|
|
436 |
$fs = get_file_storage();
|
|
|
437 |
return $fs->delete_area_files(\context_system::instance()->id, 'analytics', false, $modelid);
|
|
|
438 |
}
|
|
|
439 |
|
|
|
440 |
/**
|
|
|
441 |
* Returns the file name to be used.
|
|
|
442 |
*
|
|
|
443 |
* @param strinbool $evaluation
|
|
|
444 |
* @return string
|
|
|
445 |
*/
|
|
|
446 |
protected static function get_filename($evaluation) {
|
|
|
447 |
|
|
|
448 |
if ($evaluation === true) {
|
|
|
449 |
$filename = self::EVALUATION_FILENAME;
|
|
|
450 |
} else {
|
|
|
451 |
// Incremental time, the lock will make sure we don't have concurrency problems.
|
|
|
452 |
$filename = microtime(true) . '.csv';
|
|
|
453 |
}
|
|
|
454 |
|
|
|
455 |
return $filename;
|
|
|
456 |
}
|
|
|
457 |
}
|