edX Analytics Conference Workshop 2016

dashboard.patch

From add560e7e139e2671f6f3eb9bda5038e69114d69 Mon Sep 17 00:00:00 2001
From: Dennis Jen <djen@edx.org>
Date: Mon, 13 Jun 2016 22:28:28 -0400
Subject: [PATCH] WIP: workshop demo of course views

---
 .../courses/presenters/engagement.py               |  3 ++
 .../templates/courses/engagement_content.html      | 24 +++++++++++++++
 analytics_dashboard/courses/views/engagement.py    |  3 ++
 .../learner_analytics_api/v0/views.py              | 12 ++++----
 .../static/js/engagement-content-main.js           | 35 ++++++++++++++++++++--
 .../static/js/performance-content-main.js          |  2 +-
 requirements/base.txt                              |  2 +-
 7 files changed, 71 insertions(+), 10 deletions(-)

diff --git a/analytics_dashboard/courses/presenters/engagement.py b/analytics_dashboard/courses/presenters/engagement.py
index 191a398..3f48cec 100644
--- a/analytics_dashboard/courses/presenters/engagement.py
+++ b/analytics_dashboard/courses/presenters/engagement.py
@@ -132,6 +132,9 @@ def get_summary_and_trend_data(self):
             self._annotate_with_enrollment(summary, trends, enrollment_data)
         return summary, trends
 
+    def get_course_views(self):
+        return self.course.views()
+
 
 class CourseEngagementVideoPresenter(CourseAPIPresenterMixin, BasePresenter):
 
diff --git a/analytics_dashboard/courses/templates/courses/engagement_content.html b/analytics_dashboard/courses/templates/courses/engagement_content.html
index 9f600f5..5cf5a86 100644
--- a/analytics_dashboard/courses/templates/courses/engagement_content.html
+++ b/analytics_dashboard/courses/templates/courses/engagement_content.html
@@ -42,6 +42,30 @@ <h4 class="section-title">{% trans "Weekly Student Engagement" %}</h4>
     {% endif %}
   </section>
 
+  <section class="view-section" aria-hidden="true">
+    <div class="section-heading">
+      <h4 class="section-title">{% trans "Content Views" %}</h4>
+      <span class="section-heading-note small">{% trans "How many views?" %}</span>
+    </div>
+
+    {% if js_data.course.courseViews %}
+        <div class="section-content section-data-graph">
+          <div class="section-content section-data-viz">
+                <div class="analytics-chart-container">
+                    <div class="chart-info">{% trans "Views" %}</div>
+                    {% trans "Course Views." as tip_text %}
+                    {% include "chart_tooltip.html" with tip_text=tip_text track_category="trend"%}
+                    <div id="course-views" class="analytics-chart">
+                        {% include "loading.html" %}
+                    </div>
+                </div>
+          </div>
+        </div>
+    {% else %}
+        {% show_chart_error %}
+    {% endif %}
+  </section>
+
   <section class="view-section">
     <div class="section-heading bordered">
       <h4 class="section-title">{% trans "Student Activity Metrics" %}</h4>
diff --git a/analytics_dashboard/courses/views/engagement.py b/analytics_dashboard/courses/views/engagement.py
index 56e1257..1d71ff2 100644
--- a/analytics_dashboard/courses/views/engagement.py
+++ b/analytics_dashboard/courses/views/engagement.py
@@ -45,13 +45,16 @@ def get_context_data(self, **kwargs):
         summary = None
         trends = None
         last_updated = None
+        course_views = None
         try:
             summary, trends = self.presenter.get_summary_and_trend_data()
+            course_views = self.presenter.get_course_views()
             last_updated = summary['last_updated']
         except NotFoundError:
             logger.error("Failed to retrieve engagement content data for %s.", self.course_id)
 
         context['js_data']['course']['engagementTrends'] = trends
+        context['js_data']['course']['courseViews'] = course_views
         context.update({
             'summary': summary,
             'update_message': self.get_last_updated_message(last_updated)
diff --git a/analytics_dashboard/learner_analytics_api/v0/views.py b/analytics_dashboard/learner_analytics_api/v0/views.py
index fc94995..cdc7aeb 100644
--- a/analytics_dashboard/learner_analytics_api/v0/views.py
+++ b/analytics_dashboard/learner_analytics_api/v0/views.py
@@ -1,6 +1,6 @@
 import json
 
-from requests.exceptions import ConnectTimeout
+# from requests.exceptions import ConnectTimeout
 
 from rest_framework.exceptions import PermissionDenied
 from rest_framework.generics import RetrieveAPIView
@@ -52,11 +52,11 @@ def handle_exception(self, exc):
         Handles timeouts raised by the API client by returning an HTTP
         504.
         """
-        if isinstance(exc, ConnectTimeout):
-            return Response(
-                data={'developer_message': 'Learner Analytics API timed out.', 'error_code': 'analytics_api_timeout'},
-                status=504
-            )
+        # if isinstance(exc, ConnectTimeout):
+        #     return Response(
+        #         data={'developer_message': 'Learner Analytics API timed out.', 'error_code': 'analytics_api_timeout'},
+        #         status=504
+        #     )
         return super(BaseLearnerApiView, self).handle_exception(exc)
 
 
diff --git a/analytics_dashboard/static/js/engagement-content-main.js b/analytics_dashboard/static/js/engagement-content-main.js
index d0f9de4..9cb0738 100644
--- a/analytics_dashboard/static/js/engagement-content-main.js
+++ b/analytics_dashboard/static/js/engagement-content-main.js
@@ -6,7 +6,8 @@
 require(['vendor/domReady!', 'load/init-page'], function (doc, page) {
     'use strict';
 
-    require(['underscore', 'views/data-table-view', 'views/trends-view'], function (_, DataTableView, TrendsView) {
+    require(['underscore', 'views/data-table-view', 'views/trends-view', 'views/stacked-bar-view'],
+        function (_, DataTableView, TrendsView, StackedBarView) {
         // shared settings between the chart and table
         // colors are chosen to be color-blind accessible
         var settings = [
@@ -51,7 +52,8 @@ require(['vendor/domReady!', 'load/init-page'], function (doc, page) {
                     type: 'percent'
                 }
             ],
-            trendSettings;
+            trendSettings,
+            courseViewsColumns;
 
         // remove settings for data that doesn't exist (ex. forums)
         settings = _(settings).filter(function (setting) {
@@ -92,5 +94,34 @@ require(['vendor/domReady!', 'load/init-page'], function (doc, page) {
             sorting: ['-weekEnding'],
             replaceNull: '-'
         });
+
+        courseViewsColumns = [
+            {
+                key: 'total_views',
+                title: gettext('Average Correct'),
+                className: 'text-right',
+                type: 'number',
+                color: '#4BB4FB'
+            },
+            {
+                key: 'unique_user_views',
+                title: gettext('Average Incorrect'),
+                className: 'text-right',
+                type: 'number',
+                color: '#CA0061'
+            }
+        ];
+        // create a unique name...
+        _(page.models.courseModel.get('courseViews')).each(function (view) {
+            view.name = [view.section, view.subsection].join('_');
+        });
+
+        new StackedBarView({
+            el: '#course-views',
+            model: page.models.courseModel,
+            modelAttribute: 'courseViews',
+            trends: courseViewsColumns
+        });
+
     });
 });
diff --git a/analytics_dashboard/static/js/performance-content-main.js b/analytics_dashboard/static/js/performance-content-main.js
index 9cb4819..f36e6e8 100644
--- a/analytics_dashboard/static/js/performance-content-main.js
+++ b/analytics_dashboard/static/js/performance-content-main.js
@@ -50,7 +50,7 @@ require(['vendor/domReady!', 'load/init-page'], function (doc, page) {
 
             if (model.get('hasData')) {
                 new StackedBarView({
-                    el: '#chart-view',
+                    el: '#course-views',
                     model: model,
                     modelAttribute: 'primaryContent',
                     dataType: 'decimal',
diff --git a/requirements/base.txt b/requirements/base.txt
index 1568d14..36520ae 100644
--- a/requirements/base.txt
+++ b/requirements/base.txt
@@ -25,7 +25,7 @@ python-social-auth==0.2.14
 git+https://github.com/pinax/django-announcements.git@f85e690705e038a62407abe54ac195f60760934b#egg=django-announcements # MIT
 
 git+https://github.com/edx/django-lang-pref-middleware.git@0.1.0#egg=django-lang-pref-middleware
-git+https://github.com/edx/edx-analytics-data-api-client.git@0.6.1#egg=edx-analytics-data-api-client==0.6.1  # edX
+git+https://github.com/edx/edx-analytics-data-api-client.git@0da6968783fa795dbd770f53ce74af2780d6ca5c#egg=edx-analytics-data-api-client==0.6.1  # edX
 git+https://github.com/edx/i18n-tools.git@0d7847f9dfa2281640527b4dc51f5854f950f9b7#egg=i18n_tools
 git+https://github.com/edx/opaque-keys.git@d45d0bd8d64c69531be69178b9505b5d38806ce0#egg=opaque-keys
 # custom opaque-key implementations for ccx

data-api-client-workshop.patch

From 0da6968783fa795dbd770f53ce74af2780d6ca5c Mon Sep 17 00:00:00 2001
From: Dennis Jen <djen@edx.org>
Date: Mon, 13 Jun 2016 17:11:55 -0400
Subject: [PATCH] WIP: adding course views demo endpoint

---
 analyticsclient/course.py | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/analyticsclient/course.py b/analyticsclient/course.py
index 83f077e..5a3e2de 100644
--- a/analyticsclient/course.py
+++ b/analyticsclient/course.py
@@ -21,6 +21,11 @@ def __init__(self, client, course_id):
         self.client = client
         self.course_id = unicode(course_id)
 
+    def views(self, data_format=DF.JSON):
+        path = 'courses/{0}/views/'.format(self.course_id)
+        return self.client.get(path, data_format=data_format)
+
+
     def enrollment(self, demographic=None, start_date=None, end_date=None, data_format=DF.JSON):
         """
         Get course enrollment data.

data-api.patch

From 5f98937172c09faaa0c8501906fbe9f0c7979edb Mon Sep 17 00:00:00 2001
From: Gabe Mulley <gabe@edx.org>
Date: Mon, 13 Jun 2016 11:57:50 -0400
Subject: [PATCH 1/3] initial API attempt

---
 analytics_data_api/v0/models.py        | 11 +++++++++++
 analytics_data_api/v0/serializers.py   |  6 ++++++
 analytics_data_api/v0/urls/courses.py  |  3 ++-
 analytics_data_api/v0/views/courses.py |  9 +++++++++
 4 files changed, 28 insertions(+), 1 deletion(-)

diff --git a/analytics_data_api/v0/models.py b/analytics_data_api/v0/models.py
index ee004df..64af53d 100644
--- a/analytics_data_api/v0/models.py
+++ b/analytics_data_api/v0/models.py
@@ -497,3 +497,14 @@ class ModuleEngagementMetricRanges(models.Model):
 
     class Meta(object):
         db_table = 'module_engagement_metric_ranges'
+
+
+class ViewDistribution(models.Model):
+    course_id = models.CharField(db_index=True, max_length=255)
+    section = models.CharField(db_index=True, max_length=255)
+    subsection = models.CharField(db_index=True, max_length=255)
+    unique_user_views = models.IntegerField()
+    total_views = models.IntegerField()
+
+    class Meta(object):
+        db_table = 'content_views'
diff --git a/analytics_data_api/v0/serializers.py b/analytics_data_api/v0/serializers.py
index 93352cc..1543b5e 100644
--- a/analytics_data_api/v0/serializers.py
+++ b/analytics_data_api/v0/serializers.py
@@ -478,3 +478,9 @@ def get_engagement_ranges(self, obj):
                 })
 
         return engagement_ranges
+
+
+class ViewDistributionSerializer(ModelSerializerWithCreatedField):
+
+    class Meta(object):
+        model = models.ViewDistribution
diff --git a/analytics_data_api/v0/urls/courses.py b/analytics_data_api/v0/urls/courses.py
index d219095..6fc823d 100644
--- a/analytics_data_api/v0/urls/courses.py
+++ b/analytics_data_api/v0/urls/courses.py
@@ -14,7 +14,8 @@
     ('enrollment/location', views.CourseEnrollmentByLocationView, 'enrollment_by_location'),
     ('problems', views.ProblemsListView, 'problems'),
     ('problems_and_tags', views.ProblemsAndTagsListView, 'problems_and_tags'),
-    ('videos', views.VideosListView, 'videos')
+    ('videos', views.VideosListView, 'videos'),
+    ('views', views.ViewDistributionView, 'views')
 ]
 
 urlpatterns = []
diff --git a/analytics_data_api/v0/views/courses.py b/analytics_data_api/v0/views/courses.py
index 5b48d97..cfae79f 100644
--- a/analytics_data_api/v0/views/courses.py
+++ b/analytics_data_api/v0/views/courses.py
@@ -765,3 +765,12 @@ class VideosListView(BaseCourseView):
     def apply_date_filtering(self, queryset):
         # no date filtering for videos -- just return the queryset
         return queryset
+
+
+class ViewDistributionView(BaseCourseView):
+    slug = 'view-distribution'
+    serializer_class = serializers.ViewDistributionSerializer
+    model = models.ViewDistribution
+
+    def apply_date_filtering(self, queryset):
+        return queryset
\ No newline at end of file

From 155d04ae05dacdfe7f91b1b5dbbcc724a99cbe1c Mon Sep 17 00:00:00 2001
From: Gabe Mulley <gabe@edx.org>
Date: Mon, 13 Jun 2016 12:01:52 -0400
Subject: [PATCH 2/3] add in a created field

---
 analytics_data_api/v0/models.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/analytics_data_api/v0/models.py b/analytics_data_api/v0/models.py
index 64af53d..338401d 100644
--- a/analytics_data_api/v0/models.py
+++ b/analytics_data_api/v0/models.py
@@ -505,6 +505,7 @@ class ViewDistribution(models.Model):
     subsection = models.CharField(db_index=True, max_length=255)
     unique_user_views = models.IntegerField()
     total_views = models.IntegerField()
+    created = models.DateTimeField(auto_now_add=True)
 
     class Meta(object):
         db_table = 'content_views'

From 16acff7493c5fee8bd73280ccda6dad2765e8a53 Mon Sep 17 00:00:00 2001
From: Dennis Jen <djen@edx.org>
Date: Mon, 13 Jun 2016 22:31:55 -0400
Subject: [PATCH 3/3] WIP: generate course view data

---
 .../management/commands/generate_fake_course_data.py     | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)

diff --git a/analytics_data_api/management/commands/generate_fake_course_data.py b/analytics_data_api/management/commands/generate_fake_course_data.py
index 44e9f0b..c5c6c95 100644
--- a/analytics_data_api/management/commands/generate_fake_course_data.py
+++ b/analytics_data_api/management/commands/generate_fake_course_data.py
@@ -253,6 +253,18 @@ def generate_tags_distribution_data(self, course_id):
                     total_submissions=total_submissions, correct_submissions=correct_submissions
                 )
 
+    def generate_video_module_data(self, course_id, section_num, subsection_num):
+        unique_user_views = random.randint(1, 100)
+        total_views = unique_user_views + random.randint(1, 100)
+
+
+        models.ViewDistribution.objects.create(course_id=course_id,
+                                       section='section_{}'.format(section_num),
+                                       subsection='subsection_{}'.format(section_num),
+                                       unique_user_views=unique_user_views,
+                                       total_views=total_views)
+
+
     def handle(self, *args, **options):
         course_id = options['course_id']
         username = options['username']
@@ -274,3 +286,7 @@ def handle(self, *args, **options):
         self.generate_learner_engagement_data(course_id, username, start_date, end_date)
         self.generate_learner_engagement_range_data(course_id, start_date, end_date)
         self.generate_tags_distribution_data(course_id)
+
+        for i in range(10):
+            for j in range(10):
+                self.generate_video_module_data(course_id, i, j)

launch-tasks.sh

# HELLO WORLD TASK

# running the task
launch-task HelloWorldTask --local-scheduler --input-url /var/tmp/analytics.txt --output-url /tmp/output.txt

# running it again is a nop
launch-task HelloWorldTask --local-scheduler --input-url /var/tmp/analytics.txt --output-url /tmp/output.txt

# by default, the output determines if it should be re-run or not
launch-task HelloWorldTask --local-scheduler --input-url /var/tmp/devops.txt --output-url /tmp/output.txt

# change the output to have the task run
launch-task HelloWorldTask --local-scheduler --input-url /var/tmp/devops.txt --output-url /tmp/devops_output.txt

# you can mix and match input sources using URLs (this is an edX extension to luigi)
launch-task HelloWorldTask --local-scheduler --input-url hdfs://localhost:9000/tmp/analytics.txt --output-url /tmp/dfs_io.txt
launch-task HelloWorldTask --local-scheduler --input-url hdfs://localhost:9000/tmp/analytics.txt --output-url hdfs://localhost:9000/tmp/output.txt

# HELLO WORLD MAP REDUCE TASK

# note the output *must* be a DFS path, not a local path
launch-task HelloWorldMapReduceTask --local-scheduler --interval 2016 --output-root hdfs://localhost:9000/tmp/hello/ --n-reduce-tasks 1
hdfs dfs -cat hdfs://localhost:9000/tmp/hello/2016/*

# VIEW DISTRIBUTION AFTER CHALLENGES

launch-task ViewDistribution --local-scheduler --interval 2016-01-01-2016-06-16 --output-root hdfs://localhost:9000/tmp/content_views/ --n-reduce-tasks 1
hdfs dfs -cat hdfs://localhost:9000/tmp/content_views/2016-01-01-2016-06-16/*

# TO MYSQL

launch-task ViewDistributionMysqlTask --local-scheduler --interval 2016-01-01-2016-06-16 --output-root hdfs://localhost:9000/tmp/content_views/ --n-reduce-tasks 1

# AFTER CLICKING AROUND

launch-task ViewDistributionMysqlTask --local-scheduler --interval 2016-01-01-2016-06-17 --output-root hdfs://localhost:9000/tmp/content_views/ --n-reduce-tasks 1

workshop.py

from edx.analytics.tasks.mysql_load import MysqlInsertTask

from edx.analytics.tasks.url import get_target_from_url, url_path_join, ExternalURL
from edx.analytics.tasks.pathutil import EventLogSelectionMixin
from edx.analytics.tasks.mapreduce import MapReduceJobTask
from edx.analytics.tasks.util import eventlog
import luigi
import re
from edx.analytics.tasks.util.record import Record, StringField, IntegerField
import logging

log = logging.getLogger(__name__)


class HelloWorldTask(luigi.Task):

    input_url = luigi.Parameter()
    output_url = luigi.Parameter()

    def requires(self):
        return ExternalURL(self.input_url)

    def run(self):
        with self.input().open(mode='r') as input_file:
            with self.output().open(mode='w') as output_file:
                for line in input_file:
                    output_file.write('Hello ' + line)

    def output(self):
        return get_target_from_url(self.output_url)


class HelloWorldMapReduceTask(EventLogSelectionMixin, MapReduceJobTask):

    output_root = luigi.Parameter()

    def mapper(self, line):
        value = self.get_event_and_date_string(line)
        if value is None:
            return
        event, _date_string = value

        key = 'hello ' + event.get('event_type', 'UNKNOWN')
        yield key, 1

    def reducer(self, key, values):
        yield key, sum(values)

    def output(self):
        return get_target_from_url(url_path_join(self.output_root, self.interval.to_string()) + '/')


class ViewDistribution(EventLogSelectionMixin, MapReduceJobTask):

    SUBSECTION_ACCESSED_PATTERN = r'/courses/.*?courseware/([^/]+)/([^/]+)/.*$'

    output_root = luigi.Parameter()

    def mapper(self, line):
        value = self.get_event_and_date_string(line)
        if value is None:
            return
        event, _date_string = value

        event_type = event.get('event_type')
        if event_type is None:
            return

        # Challenge #1
        # Use the SUBSECTION_ACCESSED_PATTERN to extract the section and subsection from the URL. The first match will
        # be the section and the second will be the subsection.
        m = re.match(self.SUBSECTION_ACCESSED_PATTERN, event_type)
        if not m:
            return
        section, subsection = m.group(1, 2)

        username = event.get('username', '').strip()
        if not username:
            return

        course_id = eventlog.get_course_id(event)
        if not course_id:
            return

        # Challenge #2
        # What should this method yield?
        yield (course_id, section, subsection), (username)

    def reducer(self, key, values):
        # Challenge #2 continued
        # I recommend assigning each element of your key to a variable here: foo, bar = key
        course_id, section, subsection = key

        # Challenge 3
        # Compute the total number of views for each (section, subsection) as well as the unique number of users who
        # viewed that subsection.
        unique_usernames = set()
        total_views = 0
        for username in values:
            unique_usernames.add(username)
            total_views += 1
        unique_user_views = len(unique_usernames)

        yield ViewRecord(
            course_id=course_id,
            section=section,
            subsection=subsection,
            unique_user_views=unique_user_views,
            total_views=total_views
        ).to_string_tuple()

    def output(self):
        return get_target_from_url(url_path_join(self.output_root, self.interval.to_string()) + '/')


class ViewRecord(Record):
    course_id = StringField(length=255, nullable=False)
    section = StringField(length=255, nullable=False)
    subsection = StringField(length=255, nullable=False)
    unique_user_views = IntegerField()
    total_views = IntegerField()


class ViewDistributionMysqlTask(MysqlInsertTask):

    output_root = luigi.Parameter()
    interval = luigi.DateIntervalParameter()
    n_reduce_tasks = luigi.IntParameter(significant=False)

    @property
    def table(self):
        return 'content_views'

    @property
    def columns(self):
        return ViewRecord.get_sql_schema()

    @property
    def indexes(self):
        return [
            ('course_id',)
        ]

    @property
    def insert_source_task(self):
        return ViewDistribution(
            interval=self.interval,
            output_root=self.output_root,
            n_reduce_tasks=self.n_reduce_tasks
        )