From 98b2f06fe7a77bf832ddd21a2d024d3222893ca4 Mon Sep 17 00:00:00 2001
From: David Monllao <davidm@moodle.com>
Date: Fri, 23 Jun 2017 20:18:11 +0200
Subject: [PATCH] MDL-57791 mlbackend_php: Minimum 2 samples per target value

---
 admin/tool/models/lang/en/tool_models.php   |  2 +-
 lib/mlbackend/php/classes/processor.php     | 17 ++++++++++++++++-
 lib/mlbackend/php/lang/en/mlbackend_php.php |  3 ++-
 3 files changed, 19 insertions(+), 3 deletions(-)

diff --git a/admin/tool/models/lang/en/tool_models.php b/admin/tool/models/lang/en/tool_models.php
index 7fc8e40ff86..c520af255dc 100644
--- a/admin/tool/models/lang/en/tool_models.php
+++ b/admin/tool/models/lang/en/tool_models.php
@@ -26,7 +26,7 @@ $string['accuracy'] = 'Accuracy';
 $string['allpredictions'] = 'All predictions';
 $string['analysingsitedata'] = 'Analysing the site';
 $string['analyticmodels'] = 'Analytic models';
-$string['bettercli'] = 'To evaluate models and to get predictions are heavy processes, it is better to run them through command line interface';
+$string['bettercli'] = 'Evaluating models and generating predictions may involve heavy processing. It is advised that you run these actions via the command line interface';
 $string['cantguessstartdate'] = 'Can\'t guess the start date';
 $string['cantguessenddate'] = 'Can\'t guess the end date';
 $string['clienablemodel'] = 'You can enable the model by selecting a time splitting method by its id. Note that you can also enable it later using the web interface (\'none\' to exit)';
diff --git a/lib/mlbackend/php/classes/processor.php b/lib/mlbackend/php/classes/processor.php
index 294f8cdffd3..1981102bcb4 100644
--- a/lib/mlbackend/php/classes/processor.php
+++ b/lib/mlbackend/php/classes/processor.php
@@ -258,6 +258,21 @@ class processor implements \core_analytics\predictor {
         }
         fclose($fh);
 
+        // We need at least 2 samples belonging to each target.
+        $counts = array_count_values($targets);
+        foreach ($counts as $count) {
+            if ($count < 2) {
+                $notenoughdata = true;
+            }
+        }
+        if (!empty($notenoughdata)) {
+            $resultobj = new \stdClass();
+            $resultobj->status = \core_analytics\model::EVALUATE_NOT_ENOUGH_DATA;
+            $resultobj->score = 0;
+            $resultobj->info = array(get_string('errornotenoughdata', 'mlbackend_php'));
+            return $resultobj;
+        }
+
         $phis = array();
 
         // Evaluate the model multiple times to confirm the results are not significantly random due to a short amount of data.
@@ -318,7 +333,7 @@ class processor implements \core_analytics\predictor {
             $a = new \stdClass();
             $a->deviation = $modeldev;
             $a->accepteddeviation = $maxdeviation;
-            $resultobj->info[] = get_string('errornotenoughdata', 'mlbackend_php', $a);
+            $resultobj->info[] = get_string('errornotenoughdatadev', 'mlbackend_php', $a);
         }
 
         if ($resultobj->score < \core_analytics\model::MIN_SCORE) {
diff --git a/lib/mlbackend/php/lang/en/mlbackend_php.php b/lib/mlbackend/php/lang/en/mlbackend_php.php
index 2a821a40b36..ed0718dce6b 100644
--- a/lib/mlbackend/php/lang/en/mlbackend_php.php
+++ b/lib/mlbackend/php/lang/en/mlbackend_php.php
@@ -25,6 +25,7 @@
 $string['datasetsizelimited'] = 'Only a part of the evaluation dataset has been evaluated due to its size. Set $CFG->mlbackend_php_no_memory_limit if you are confident that your system can cope a {$a} dataset';
 $string['errorcantloadmodel'] = 'Model file {$a} does not exist, ensure the model has been trained before using it to predict.';
 $string['errorlowscore'] = 'The evaluated model prediction accuracy is not very high, some predictions may not be accurate. Model score = {$a->score}, minimum score = {$a->minscore}';
-$string['errornotenoughdata'] = 'The evaluation results varied too much, you could try to gather more data to ensure the model is valid. Evaluation results standard deviation = {$a->deviation}, maximum recommended standard deviation = {$a->accepteddeviation}';
+$string['errornotenoughdata'] = 'There is no enough data to evaluate this model using this time splitting method.';
+$string['errornotenoughdatadev'] = 'The evaluation results varied too much, you could try to gather more data to ensure the model is valid. Evaluation results standard deviation = {$a->deviation}, maximum recommended standard deviation = {$a->accepteddeviation}';
 $string['errorphp7required'] = 'The PHP machine learning backend requires PHP 7';
 $string['pluginname'] = 'PHP machine learning backend';