From 8b5a05ba06a285a4ea0accc2cf64e2f87ca4e1eb Mon Sep 17 00:00:00 2001
From: Maximilian Grundke <maximiliangrundke@googlemail.com>
Date: Wed, 22 Nov 2017 17:40:14 +0100
Subject: [PATCH] Detect exercises with too high or too low working time
 average

---
 lib/tasks/detect_exercise_anomalies.rake | 44 ++++++++++++++++++++++++
 1 file changed, 44 insertions(+)
 create mode 100644 lib/tasks/detect_exercise_anomalies.rake

diff --git a/lib/tasks/detect_exercise_anomalies.rake b/lib/tasks/detect_exercise_anomalies.rake
new file mode 100644
index 00000000..d0be6795
--- /dev/null
+++ b/lib/tasks/detect_exercise_anomalies.rake
@@ -0,0 +1,44 @@
+namespace :detect_exercise_anomalies do
+
+  task :with_at_least, [:number_of_exercises, :number_of_solutions] => :environment do |task, args|
+    number_of_exercises = args.number_of_exercises
+    number_of_solutions = args.number_of_solutions
+
+    # These factors determine if an exercise is an anomaly, given the average working time (avg):
+    # (avg * MIN_TIME_FACTOR) <= working_time <= (avg * MAX_TIME_FACTOR)
+    MIN_TIME_FACTOR = 0.1
+    MAX_TIME_FACTOR = 2
+
+    # Get all exercise collections that have at least the specified amount of exercises and at least the specified
+    # number of submissions AND are flagged for anomaly detection
+    collections = ExerciseCollection
+                      .where(:use_anomaly_detection => true)
+                      .joins("join exercise_collections_exercises ece on exercise_collections.id = ece.exercise_collection_id
+                              join
+                                (select e.id
+                                 from exercises e
+                                   join submissions s on s.exercise_id = e.id
+                                 group by e.id
+                                 having count(s.id) > #{ExerciseCollection.sanitize(number_of_solutions)}
+                                ) as exercises_with_submissions on exercises_with_submissions.id = ece.exercise_id")
+                      .group('exercise_collections.id')
+                      .having('count(exercises_with_submissions.id) > ?', number_of_exercises)
+
+    collections.each do |collection|
+      puts "\t- #{collection}"
+      working_times = {}
+      collection.exercises.each do |exercise|
+        puts "\t\t> #{exercise.title}"
+        avgwt = exercise.average_working_time.split(':')
+        seconds = avgwt[0].to_i * 60 * 60 + avgwt[1].to_i * 60 + avgwt[2].to_f
+        working_times[exercise.id] = seconds
+      end
+      average = working_times.values.reduce(:+) / working_times.size
+      anomalies = working_times.select do |exercise_id, working_time|
+        working_time > average * MAX_TIME_FACTOR or working_time < average * MIN_TIME_FACTOR
+      end
+      puts anomalies
+    end
+  end
+
+end