From 474e5c539d96158d57f1171f541d76ee588117ab Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?David=20Monlla=C3=B3?= <davidm@moodle.com>
Date: Tue, 21 May 2019 16:56:25 +0200
Subject: [PATCH] MDL-65634 analytics: Discard late and early student
 enrolments

---
 .../analytics/target/course_dropout.php       | 38 +++++++++++++++++++
 1 file changed, 38 insertions(+)

diff --git a/lib/classes/analytics/target/course_dropout.php b/lib/classes/analytics/target/course_dropout.php
index b44ec4d34ae..4cd82752828 100644
--- a/lib/classes/analytics/target/course_dropout.php
+++ b/lib/classes/analytics/target/course_dropout.php
@@ -203,6 +203,8 @@ class course_dropout extends \core_analytics\local\target\binary {
      */
     public function is_valid_sample($sampleid, \core_analytics\analysable $course, $fortraining = true) {
 
+        $now = time();
+
         $userenrol = $this->retrieve('user_enrolments', $sampleid);
         if ($userenrol->timeend && $course->get_start() > $userenrol->timeend) {
             // Discard enrolments which time end is prior to the course start. This should get rid of
@@ -227,6 +229,16 @@ class course_dropout extends \core_analytics\local\target\binary {
             return false;
         }
 
+        if ($now < $userenrol->timestart && $userenrol->timestart) {
+            // Discard enrolments whose start date is after now (no need to check timecreated > $now :P).
+            return false;
+        }
+
+        if (!$fortraining && $userenrol->timeend && $userenrol->timeend < $now) {
+            // We don't want to generate predictions for finished enrolments.
+            return false;
+        }
+
         return true;
     }
 
@@ -245,6 +257,11 @@ class course_dropout extends \core_analytics\local\target\binary {
      */
     protected function calculate_sample($sampleid, \core_analytics\analysable $course, $starttime = false, $endtime = false) {
 
+        if ($this->enrolment_starts_after_calculation_start($sampleid, $starttime)) {
+            // Discard user enrolments whose start date is after $starttime.
+            return null;
+        }
+
         $userenrol = $this->retrieve('user_enrolments', $sampleid);
 
         // We use completion as a success metric only when it is enabled.
@@ -273,4 +290,25 @@ class course_dropout extends \core_analytics\local\target\binary {
         }
         return 0;
     }
+
+    /**
+     * Does the user enrolment created after this time range start time or starts after it?
+     *
+     * We need to identify these enrolments because the indicators can not be calculated properly
+     * if the student enrolment started half way through this time range.
+     *
+     * User enrolments whose end date is before time() have already been discarded in
+     * course_enrolments::is_valid_sample.
+     *
+     * @param  int    $sampleid
+     * @param  int    $starttime
+     * @return bool
+     */
+    protected function enrolment_starts_after_calculation_start(int $sampleid, int $starttime) {
+        $userenrol = $this->retrieve('user_enrolments', $sampleid);
+        if ($userenrol->timestart && $userenrol->timestart > $starttime) {
+            return true;
+        }
+        return false;
+    }
 }