Merge branch 'master' into jhawk0224/dock-clear-button

.github/workflows/shared-build-deploy.yaml

@@ -52,7 +52,7 @@ jobs:
     strategy:
       fail-fast: true
       matrix:
-        product: [review, plan, alert]
+        product: [review, plan, alert, degree-plan]
     with:
       path: frontend/${{ matrix.product }}
       nodeVersion: 18.19.0
@@ -62,7 +62,7 @@ jobs:
     strategy:
       fail-fast: true
       matrix:
-        product: [{name: pcr, path: review}, {name: pcp, path: plan}, {name: pca, path: alert}]
+        product: [{name: pcr, path: review}, {name: pcp, path: plan}, {name: pca, path: alert}, {name: pdp, path: degree-plan}]
     with:
       # Inputs
       imageName: "${{ matrix.product.name }}-frontend"

backend/courses/management/commands/recompute_soft_state.py

@@ -99,7 +99,9 @@ def recompute_enrollment():
         )
 
 
-# course credits = sum(section credis for all activities)
+# course credits = sum(section credis for all activities for sections below 500)
+# the < 500 heuristic comes from here:
+# https://provider.www.upenn.edu/computing/da/dw/student/enrollment_section_type.e.html
 COURSE_CREDITS_RAW_SQL = dedent(
     """
     WITH CourseCredits AS (
@@ -108,6 +110,7 @@ def recompute_enrollment():
         INNER JOIN (
             SELECT MAX(U1."credits") AS "activity_cus", U1."course_id"
             FROM "courses_section" U1
+            WHERE U1."code" < '500' AND (U1."status" <> 'X' OR U1."status" <> '')
             GROUP BY U1."course_id", U1."activity"
         ) AS U2
         ON U0."id" = U2."course_id"
@@ -125,7 +128,6 @@ def recompute_enrollment():
 def recompute_course_credits(
     model=Course,  # so this function can be used in migrations (see django.db.migrations.RunPython)
 ):
-
     with connection.cursor() as cursor:
         cursor.execute(COURSE_CREDITS_RAW_SQL)
 

backend/courses/management/commands/recompute_topics.py

@@ -1,9 +1,10 @@
 from django.core.management.base import BaseCommand
 from django.db import transaction
 from django.db.models import Count, OuterRef, Subquery
+from tqdm import tqdm
 
 from courses.models import Course, Topic
-from courses.util import all_semesters
+from courses.util import all_semesters, historical_semester_probability
 
 
 def garbage_collect_topics():
@@ -151,5 +152,28 @@ def handle(self, *args, **kwargs):
             assert (
                 min_semester in all_semesters()
             ), f"--min-semester={min_semester} is not a valid semester."
-
+        semesters = sorted(
+            [sem for sem in all_semesters() if not min_semester or sem >= min_semester]
+        )
         recompute_topics(min_semester, verbose=True, allow_null_parent_topic=bool(min_semester))
+        recompute_historical_semester_probabilities(current_semester=semesters[-1], verbose=True)
+
+
+def recompute_historical_semester_probabilities(current_semester, verbose=False):
+    """
+    Recomputes the historical probabilities for all topics.
+    """
+    if verbose:
+        print("Recomputing historical probabilities for all topics...")
+    topics = Topic.objects.all()
+    # Iterate over each Topic
+    for i, topic in tqdm(enumerate(topics), disable=not verbose, total=topics.count()):
+        # Calculate historical_year_probability for the current topic
+        ordered_courses = topic.courses.all().order_by("semester")
+        ordered_semester = [course.semester for course in ordered_courses]
+        historical_prob = historical_semester_probability(current_semester, ordered_semester)
+        # Update the historical_probabilities field for the current topic
+        topic.historical_probabilities_spring = historical_prob[0]
+        topic.historical_probabilities_summer = historical_prob[1]
+        topic.historical_probabilities_fall = historical_prob[2]
+        topic.save()

backend/courses/migrations/0064_auto_20240225_1331.py

@@ -0,0 +1,66 @@
+# Generated by Django 3.2.23 on 2024-02-25 18:31
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ("courses", "0063_auto_20231212_1750"),
+    ]
+
+    operations = [
+        migrations.AddField(
+            model_name="topic",
+            name="historical_probabilities_fall",
+            field=models.FloatField(
+                default=0,
+                help_text="\nThe historical probability of a student taking a course in this topic in the fall\nsemester, based on historical data. This field is recomputed nightly from the\n`parent_course` graph (in the recompute_soft_state cron job).\n",
+            ),
+        ),
+        migrations.AddField(
+            model_name="topic",
+            name="historical_probabilities_spring",
+            field=models.FloatField(
+                default=0,
+                help_text="\nThe historical probability of a student taking a course in this topic in the spring\nsemester, based on historical data. This field is recomputed nightly from the\n`parent_course` graph (in the recompute_soft_state cron job).\n",
+            ),
+        ),
+        migrations.AddField(
+            model_name="topic",
+            name="historical_probabilities_summer",
+            field=models.FloatField(
+                default=0,
+                help_text="\nThe historical probability of a student taking a course in this topic in the summer\nsemester, based on historical data. This field is recomputed nightly from the\n`parent_course` graph (in the recompute_soft_state cron job).\n",
+            ),
+        ),
+        migrations.AlterField(
+            model_name="section",
+            name="activity",
+            field=models.CharField(
+                choices=[
+                    ("", "Undefined"),
+                    ("CLN", "Clinic"),
+                    ("CRT", "Clinical Rotation"),
+                    ("DAB", "Dissertation Abroad"),
+                    ("DIS", "Dissertation"),
+                    ("DPC", "Doctoral Program Exchange"),
+                    ("FLD", "Field Work"),
+                    ("HYB", "Hybrid"),
+                    ("IND", "Independent Study"),
+                    ("LAB", "Lab"),
+                    ("LEC", "Lecture"),
+                    ("MST", "Masters Thesis"),
+                    ("ONL", "Online"),
+                    ("PRC", "Practicum"),
+                    ("REC", "Recitation"),
+                    ("SEM", "Seminar"),
+                    ("SRT", "Senior Thesis"),
+                    ("STU", "Studio"),
+                ],
+                db_index=True,
+                help_text='The section activity, e.g. `LEC` for CIS-120-001 (2020A). Options and meanings: <table width=100%><tr><td>""</td><td>"Undefined"</td></tr><tr><td>"CLN"</td><td>"Clinic"</td></tr><tr><td>"CRT"</td><td>"Clinical Rotation"</td></tr><tr><td>"DAB"</td><td>"Dissertation Abroad"</td></tr><tr><td>"DIS"</td><td>"Dissertation"</td></tr><tr><td>"DPC"</td><td>"Doctoral Program Exchange"</td></tr><tr><td>"FLD"</td><td>"Field Work"</td></tr><tr><td>"HYB"</td><td>"Hybrid"</td></tr><tr><td>"IND"</td><td>"Independent Study"</td></tr><tr><td>"LAB"</td><td>"Lab"</td></tr><tr><td>"LEC"</td><td>"Lecture"</td></tr><tr><td>"MST"</td><td>"Masters Thesis"</td></tr><tr><td>"ONL"</td><td>"Online"</td></tr><tr><td>"PRC"</td><td>"Practicum"</td></tr><tr><td>"REC"</td><td>"Recitation"</td></tr><tr><td>"SEM"</td><td>"Seminar"</td></tr><tr><td>"SRT"</td><td>"Senior Thesis"</td></tr><tr><td>"STU"</td><td>"Studio"</td></tr></table>',
+                max_length=50,
+            ),
+        ),
+    ]

backend/courses/models.py

@@ -377,6 +377,36 @@ class Topic(models.Model):
         ),
     )
 
+    historical_probabilities_spring = models.FloatField(
+        default=0,
+        help_text=dedent(
+            """
+        The historical probability of a student taking a course in this topic in the spring
+        semester, based on historical data. This field is recomputed nightly from the
+        `parent_course` graph (in the recompute_soft_state cron job).
+        """
+        ),
+    )
+    historical_probabilities_summer = models.FloatField(
+        default=0,
+        help_text=dedent(
+            """
+        The historical probability of a student taking a course in this topic in the summer
+        semester, based on historical data. This field is recomputed nightly from the
+        `parent_course` graph (in the recompute_soft_state cron job).
+        """
+        ),
+    )
+    historical_probabilities_fall = models.FloatField(
+        default=0,
+        help_text=dedent(
+            """
+        The historical probability of a student taking a course in this topic in the fall
+        semester, based on historical data. This field is recomputed nightly from the
+        `parent_course` graph (in the recompute_soft_state cron job).
+        """
+        ),
+    )
     branched_from = models.ForeignKey(
         "Topic",
         related_name="branched_to",

backend/courses/serializers.py

@@ -15,6 +15,7 @@
     StatusUpdate,
     UserProfile,
 )
+from plan.management.commands.recommendcourses import cosine_similarity
 
 
 class MeetingSerializer(serializers.ModelSerializer):
@@ -251,7 +252,23 @@ def get_num_sections(self, obj):
         return obj.sections.count()
 
     def get_recommendation_score(self, obj):
-        return 0
+        user_vector = self.context.get("user_vector")
+        curr_course_vectors_dict = self.context.get("curr_course_vectors_dict")
+
+        if user_vector is None or curr_course_vectors_dict is None:
+            # NOTE: there should be no case in which user_vector is None
+            # but curr_course_vectors_dict is not None. However, for
+            # stability in production, recommendation_score is None when
+            # either is None
+            return None
+
+        course_vector = curr_course_vectors_dict.get(obj.full_code)
+        if course_vector is None:
+            # Fires when the curr_course_vectors_dict is defined (ie, the user is authenticated)
+            # but the course code is not in the model
+            return None
+
+        return cosine_similarity(course_vector, user_vector)
 
     course_quality = serializers.DecimalField(
         max_digits=4, decimal_places=3, read_only=True, help_text=course_quality_help

backend/courses/util.py

@@ -721,3 +721,53 @@ def get_semesters(semesters: str = None) -> list[str]:
             if s not in possible_semesters:
                 raise ValueError(f"Provided semester {s} was not found in the db.")
     return sorted(semesters)
+
+
+def historical_semester_probability(current_semester: str, semesters: list[str]):
+    """
+    :param current: The current semester represented in the 20XX(A|B|C) format.
+    :type current: str
+    :param courses: A list of Course objects sorted by date in ascending order.
+    :type courses: list
+    :returns: A list of 3 probabilities representing the likelihood of
+    taking a course in each semester.
+    :rtype: list
+    """
+    PROB_DISTRIBUTION = [0.4, 0.3, 0.15, 0.1, 0.05]
+
+    def normalize_and_round(prob, i):
+        """Modifies the probability distribution to account for the
+        fact that the last course was taken i years ago."""
+        truncate = PROB_DISTRIBUTION[:i]
+        total = sum(truncate)
+        return list(map(lambda x: round(x / total, 3), truncate))
+
+    semester_probabilities = {"A": 0.0, "B": 0.0, "C": 0.0}
+    current_year = int(current_semester[:-1])
+    semesters = [
+        semester
+        for semester in semesters
+        if semester < str(current_year) and semester > str(current_year - 5)
+    ]
+    if not semesters:
+        return [0, 0, 0]
+    if current_year - int(semesters[0][:-1]) < 5:
+        # If the class hasn't been offered in the last 5 years,
+        # we make sure the resulting probabilities sum to 1
+        modified_prob_distribution = normalize_and_round(
+            PROB_DISTRIBUTION, current_year - int(semesters[0][:-1])
+        )
+    else:
+        modified_prob_distribution = PROB_DISTRIBUTION
+    for historical_semester in semesters:
+        historical_year = int(historical_semester[:-1])
+        sem_char = historical_semester[-1].upper()  # A, B, C
+        semester_probabilities[sem_char] += modified_prob_distribution[
+            current_year - historical_year - 1
+        ]
+    return list(
+        map(
+            lambda x: min(round(x, 2), 1.00),
+            [semester_probabilities["A"], semester_probabilities["B"], semester_probabilities["C"]],
+        )
+    )

backend/courses/views.py

@@ -36,6 +36,7 @@
 )
 from courses.util import get_current_semester
 from PennCourses.docs_settings import PcxAutoSchema
+from plan.management.commands.recommendcourses import retrieve_course_clusters, vectorize_user
 
 
 SEMESTER_PARAM_DESCRIPTION = (
@@ -66,8 +67,12 @@ def filter_by_semester(self, queryset):
         semester = self.get_semester()
         if semester != "all":
             queryset = queryset.filter(**{self.get_semester_field(): semester})
-        else:
-            queryset = queryset.order_by("full_code", "-semester").distinct("full_code")
+        else:  # Only used for Penn Degree Plan (as of 4/10/2024)
+            queryset = (
+                queryset.exclude(credits=None)  # heuristic: if the credits are empty, then ignore
+                .order_by("full_code", "-semester")
+                .distinct("full_code")
+            )
         return queryset
 
     def get_queryset(self):
@@ -209,6 +214,15 @@ def get_serializer_context(self):
 
         if self.request is None or not self.request.user or not self.request.user.is_authenticated:
             return context
+
+        _, _, curr_course_vectors_dict, past_course_vectors_dict = retrieve_course_clusters()
+        user_vector, _ = vectorize_user(
+            self.request.user, curr_course_vectors_dict, past_course_vectors_dict
+        )
+        context.update(
+            {"user_vector": user_vector, "curr_course_vectors_dict": curr_course_vectors_dict}
+        )
+
         return context
 
     filter_backends = [TypedCourseSearchBackend, CourseSearchFilterBackend]

backend/degree/admin.py

@@ -1,7 +1,6 @@
-from django.conf.urls import url
 from django.contrib import admin
 from django.template.response import TemplateResponse
-from django.urls import reverse
+from django.urls import re_path, reverse
 from django.utils.html import format_html
 
 from degree.models import (
@@ -16,34 +15,25 @@
 
 
 # Register your models here.
-@admin.register(Rule)
 class RuleAdmin(admin.ModelAdmin):
     search_fields = ["title", "id"]
     list_display = ["title", "id", "parent"]
     list_select_related = ["parent"]
 
 
-admin.site.register(DegreePlan)
-admin.site.register(SatisfactionStatus)
-
-
-@admin.register(PDPBetaUser)
 class PDPBetaUserAdmin(admin.ModelAdmin):
     search_fields = ("person__username", "person__id")
     autocomplete_fields = ("person",)
 
 
-@admin.register(Fulfillment)
 class FulfillmentAdmin(admin.ModelAdmin):
     autocomplete_fields = ["rules"]
 
 
-@admin.register(DoubleCountRestriction)
 class DoubleCountRestrictionAdmin(admin.ModelAdmin):
     autocomplete_fields = ["rule", "other_rule"]
 
 
-@admin.register(Degree)
 class DegreeAdmin(admin.ModelAdmin):
     autocomplete_fields = ["rules"]
     list_display = ["program", "degree", "major", "concentration", "year", "view_degree_editor"]
@@ -59,7 +49,7 @@ def get_urls(self):
         # get the default urls
         urls = super().get_urls()
         custom_urls = [
-            url(
+            re_path(
                 r"^degree-editor/$",
                 self.admin_site.admin_view(self.degree_editor),
                 name="degree-editor",
@@ -70,3 +60,12 @@ def get_urls(self):
     def degree_editor(self, request):
         context = dict(self.admin_site.each_context(request))
         return TemplateResponse(request, "degree-editor.html", context)
+
+
+admin.site.register(Rule, RuleAdmin)
+admin.site.register(DegreePlan)
+admin.site.register(SatisfactionStatus)
+admin.site.register(PDPBetaUser, PDPBetaUserAdmin)
+admin.site.register(Fulfillment, FulfillmentAdmin)
+admin.site.register(DoubleCountRestriction, DoubleCountRestrictionAdmin)
+admin.site.register(Degree, DegreeAdmin)