work on tests for score calculation for groups

mondey_backend/src/mondey_backend/models/milestones.py

@@ -193,7 +193,8 @@ class MilestoneAnswerSessionPublic(SQLModel):
     answers: dict[int, MilestoneAnswerPublic]
 
 
-# models for statistics
+# models for statistics. README: Perhaps this could be made simpler if the data was stored in a database with array-column support. sqlite apparently doesnt: https://stackoverflow.com/questions/3005231/how-to-store-array-in-one-column-in-sqlite3, but postgres does: https://www.postgresql.org/docs/9.1/arrays.html
+# will be returned to later. Issue no.
 class MilestoneAgeScore(SQLModel, table=True):
     milestone_id: int | None = Field(
         default=None,

mondey_backend/src/mondey_backend/routers/scores.py

@@ -88,31 +88,45 @@ def compute_milestonegroup_feedback_summary(
     session: SessionDep, child_id: int, answersession_id: int
 ) -> dict[int, int]:
     answersession = session.get(MilestoneAnswerSession, answersession_id)
+    print("answersession in stat: ", answersession)
     # get child age
     child = session.get(Child, child_id)
     age = get_child_age_in_months(child, answersession.created_at)
 
     # extract milestonegroups
     groups = set(answer.milestone_group_id for answer in answersession.answers.values())
-    today = datetime.today()
+    today = datetime.now()
 
     # for each milestonegroup, get the statistics, compute the current mean, and compute the feedback
     feedback: dict[int, int] = {}
     print("age: ", age)
     for group in groups:
+        print(" current group ", group, "of: ", groups)
         # try to get statistics to use as evaluation basis and recompute if it's not there
         # or is too old
         stats = session.exec(
             select(MilestoneGroupAgeScoreCollection).where(
-                MilestoneGroupAgeScoreCollection.milestonegroup_id == group
+                MilestoneGroupAgeScoreCollection.milestone_group_id == group
             )
         ).first()
-        print("stats before ", stats, stats.scores[age])
-        if stats is None or stats.created_at < today - timedelta(days=7):
-            stats = calculate_milestonegroup_statistics_by_age(session, group)
-            session.add(stats)
 
-        print("stats after ", stats, stats.scores[age])
+        print(" stats before ", stats)
+        print(stats.scores[age])
+        print(
+            " conditions: ", stats is None, stats.created_at < today - timedelta(days=7)
+        )
+        print(" created at: ", stats.created_at)
+        if stats is None or stats.created_at < today - timedelta(days=7):
+            print("recomputing stats")
+            new_stats = calculate_milestonegroup_statistics_by_age(session, group)
+            # update stuff in database
+            for new_score in new_stats.scores:
+                print(new_score)
+                session.merge(new_score)
+            session.merge(new_stats)
+            session.commit()
+            stats = new_stats
+        print(" stats after ", stats)
 
         # extract the answers for the current milestone group
         group_answers = [
@@ -122,15 +136,17 @@ def compute_milestonegroup_feedback_summary(
         ]
 
         print(
-            "group answers: ", group_answers, np.mean(group_answers), min(group_answers)
+            " group answers: ",
+            group_answers,
+            np.mean(group_answers),
+            min(group_answers),
         )
 
         # use the statistics recorded for a certain age as the basis for the feedback computation
         feedback[group] = compute_feedback_simple(
             stats.scores[age], np.mean(group_answers), min(group_answers)
         )
-
-    session.commit()
+    print("feedback: ", feedback)
     return feedback
 
 

mondey_backend/src/mondey_backend/routers/statistics.py

@@ -118,6 +118,7 @@ def calculate_milestone_statistics_by_age(
             MilestoneAnswer.milestone_id == milestone_id
         )
     else:
+        # fetch all answers that have been added in answersessions after the last statistics were calculated
         answers_query = (
             select(MilestoneAnswer)
             .join(
@@ -130,29 +131,31 @@ def calculate_milestone_statistics_by_age(
             )
         )
     answers = session.exec(answers_query).all()
-
-    count, avg_scores, stddev_scores = _get_statistics_by_age(
-        answers, child_ages, count=count, avg=avg_scores, stddev=stddev_scores
-    )
-    expected_age = _get_expected_age_from_scores(avg_scores)
-
-    # overwrite last_statistics with updated stuff
-    return MilestoneAgeScoreCollection(
-        milestone_id=milestone_id,
-        expected_age=expected_age,
-        created_at=datetime.datetime.now(),
-        scores=[
-            MilestoneAgeScore(
-                age=age,
-                milestone_id=milestone_id,
-                count=count[age],
-                avg_score=avg_scores[age],
-                stddev_score=stddev_scores[age],
-                expected_score=4 if age >= expected_age else 1,
-            )
-            for age in range(0, len(avg_scores))
-        ],
-    )
+    if len(answers) == 0:
+        return last_statistics
+    else:
+        count, avg_scores, stddev_scores = _get_statistics_by_age(
+            answers, child_ages, count=count, avg=avg_scores, stddev=stddev_scores
+        )
+        expected_age = _get_expected_age_from_scores(avg_scores)
+
+        # overwrite last_statistics with updated stuff
+        return MilestoneAgeScoreCollection(
+            milestone_id=milestone_id,
+            expected_age=expected_age,
+            created_at=datetime.datetime.now(),
+            scores=[
+                MilestoneAgeScore(
+                    age=age,
+                    milestone_id=milestone_id,
+                    count=count[age],
+                    avg_score=avg_scores[age],
+                    stddev_score=stddev_scores[age],
+                    expected_score=4 if age >= expected_age else 1,
+                )
+                for age in range(0, len(avg_scores))
+            ],
+        )
 
 
 def calculate_milestonegroup_statistics_by_age(
@@ -173,8 +176,12 @@ def calculate_milestonegroup_statistics_by_age(
     avg_scores = None
     stddev_scores = None
     if last_statistics is not None:
-        count = np.array([score.count for score in last_statistics.scores])
-        avg_scores = np.array([score.avg_score for score in last_statistics.scores])
+        count = np.array(
+            [score.count for score in last_statistics.scores], dtype=np.int32
+        )
+        avg_scores = np.array(
+            [score.avg_score for score in last_statistics.scores], dtype=np.float64
+        )
         stddev_scores = np.array(
             [score.stddev_score for score in last_statistics.scores]
         )
@@ -186,6 +193,7 @@ def calculate_milestonegroup_statistics_by_age(
             col(MilestoneAnswer.milestone_group_id) == milestonegroup_id
         )
     else:
+        # fetch all answers that have been added in answersessions after the last statistics were calculated
         answer_query = (
             select(MilestoneAnswer)
             .join(
@@ -199,21 +207,40 @@ def calculate_milestonegroup_statistics_by_age(
         )
 
     answers = session.exec(answer_query).all()
-
-    count, avg, stddev = _get_statistics_by_age(
-        answers, child_ages, count=count, avg=avg_scores, stddev=stddev_scores
-    )
-    return MilestoneGroupAgeScoreCollection(
-        milestone_group_id=milestonegroup_id,
-        scores=[
-            MilestoneGroupAgeScore(
-                milestone_group_id=milestonegroup_id,
-                age=age,
-                count=count[age],
-                avg_score=avg[age],
-                stddev_score=stddev[age],
-            )
-            for age in range(0, len(avg))
-        ],
-        created_at=datetime.datetime.now(),
-    )
+    if len(answers) == 0:
+        return last_statistics
+    else:
+        print("answers for group statistics: ", answers)
+        print(
+            "old statistics: ",
+            avg_scores[avg_scores > 0],
+            stddev_scores[stddev_scores > 0],
+        )
+        count, avg, stddev = _get_statistics_by_age(
+            answers, child_ages, count=count, avg=avg_scores, stddev=stddev_scores
+        )
+        print(
+            "new statistics: ",
+            np.nonzero(count),
+            np.nonzero(avg),
+            np.nonzero(stddev),
+            count[count > 0],
+            avg[avg > 0],
+            stddev[stddev > 0],
+        )
+        return MilestoneGroupAgeScoreCollection(
+            milestone_group_id=milestonegroup_id,
+            scores=[
+                MilestoneGroupAgeScore(
+                    milestone_group_id=milestonegroup_id,
+                    age=age,
+                    count=int(
+                        count[age]
+                    ),  # need a conversion to avoid numpy.int32 being stored as byte object
+                    avg_score=avg[age],
+                    stddev_score=stddev[age],
+                )
+                for age in range(0, len(avg))
+            ],
+            created_at=datetime.datetime.now(),
+        )

mondey_backend/tests/conftest.py

@@ -270,6 +270,7 @@ def session(children: list[dict]):
                 answer_session_id=2, milestone_id=2, milestone_group_id=1, answer=2
             )
         )
+
         # add an (expired) milestone answer session for child 3 / admin user (id 1) with 1 answer
         session.add(
             MilestoneAnswerSession(
@@ -353,26 +354,43 @@ def session(children: list[dict]):
                 created_at=datetime.datetime.today() - datetime.timedelta(days=1),
             )
         )
+
         session.add(
             MilestoneAgeScoreCollection(
                 milestone_id=7,
                 expected_age=42,
-                created_at=datetime.datetime.today() - datetime.timedelta(days=1),
+                created_at=datetime.datetime.today() - datetime.timedelta(days=12),
             )
         )
-
-        # add basic statistics for milestonegroup 1
+        # add basic statistics for milestonegroups, one that is current and one that is outdated
         session.add(
             MilestoneGroupAgeScoreCollection(
                 milestonegroup_id=1,
                 created_at=datetime.datetime.today() - datetime.timedelta(days=1),
             )
         )
-
         session.add(
             MilestoneGroupAgeScoreCollection(
                 milestonegroup_id=2,
-                created_at=datetime.datetime.today() - datetime.timedelta(days=1),
+                created_at=datetime.datetime.today() - datetime.timedelta(days=12),
+            )
+        )
+
+        # add an expired milestone answer session that is newer than the last statistics for milestonegroup
+        session.add(
+            MilestoneAnswerSession(
+                child_id=3, user_id=1, created_at=today - datetime.timedelta(days=1)
+            )
+        )
+        # add two milestone answers
+        session.add(
+            MilestoneAnswer(
+                answer_session_id=4, milestone_id=7, milestone_group_id=2, answer=2
+            )
+        )
+        session.add(
+            MilestoneAnswer(
+                answer_session_id=4, milestone_id=2, milestone_group_id=1, answer=3
             )
         )
 

mondey_backend/tests/utils/test_scores.py

@@ -1,6 +1,12 @@
+from datetime import datetime
+from datetime import timedelta
+
+from sqlmodel import select
+
 from mondey_backend.models.milestones import MilestoneAgeScore
 from mondey_backend.models.milestones import MilestoneAnswerSession
 from mondey_backend.models.milestones import MilestoneGroupAgeScore
+from mondey_backend.models.milestones import MilestoneGroupAgeScoreCollection
 from mondey_backend.routers.scores import TrafficLight
 from mondey_backend.routers.scores import compute_feedback_simple
 from mondey_backend.routers.scores import compute_milestonegroup_feedback_summary
@@ -57,20 +63,51 @@ def test_compute_summary_milestonegroup_feedback_for_answersession(session):
     assert feedback[1] == TrafficLight.yellowWithCaveat.value
     assert len(feedback) == 1
 
-    # # same as above, but for answers 4, 3  -> 3.5 ==> green
-    # feedback = compute_milestonegroup_feedback_summary(
-    #     session,
-    #     child_id=1,
-    #     answersession_id=2
-    # )
-    # assert len(feedback) == 1
-    # assert feedback[1] == TrafficLight.green.value
+    # same as above, but for answers 4, 3  -> 3.5 ==> green
+    feedback = compute_milestonegroup_feedback_summary(
+        session, child_id=1, answersession_id=2
+    )
+    assert len(feedback) == 1
+    assert feedback[1] == TrafficLight.green.value
 
 
 def test_compute_summary_milestonegroup_feedback_for_answersession_no_statistics(
     session,
 ):
-    assert 5 == 9
+    feedback = compute_milestonegroup_feedback_summary(
+        session, child_id=3, answersession_id=3
+    )
+
+    # child is 42 months old, for which there is no data. Hence the feedback is invalid
+    assert len(feedback) == 1
+    assert feedback[2] == TrafficLight.invalid.value
+
+    # check that the statistics have been updated
+    statistics = session.exec(
+        select(MilestoneGroupAgeScoreCollection).where(
+            MilestoneGroupAgeScoreCollection.milestone_group_id == 2
+        )
+    ).all()
+    assert len(statistics) == 1
+    assert statistics[0].created_at >= datetime.now() - timedelta(
+        minutes=3
+    )  # can be at max 3 min old
+
+    for i, score in enumerate(statistics[0].scores):
+        if i in [8, 9]:
+            assert score.avg_score > 0
+            assert score.stddev_score > 0
+            assert score.count == 2
+
+        if i == 65:
+            assert score.avg_score > 0
+            assert score.stddev_score == 0
+            assert score.count == 1
+
+        if i not in [8, 9, 65]:
+            assert score.avg_score == 0
+            assert score.stddev_score == 0
+            assert score.count == 0
 
 
 # def test_compute_detailed_milestonegroup_feedback_for_answersession(session):