feat: New models

models/_docs/docs.md

@@ -124,6 +124,10 @@ The section, subsection, unit, and part number of the problem block. In the form
 Boolean indicating whether the responses were correct
 {% enddocs %}
 
+{% docs problem_id %}
+Unique identifier for the problem
+{% enddocs %}
+
 {% docs object_tag_source %}
 Course objects and their associated tags from CMS events:  
 `COURSE_CREATED, XBLOCK_CREATED, LIBRARY_BLOCK_CREATED, CONTENT_OBJECT_ASSOCIATIONS_CHANGED`  

models/navigation/fact_navigation_completion.sql

@@ -11,7 +11,9 @@ select
     pages.item_count as page_count,
     pages.section_with_name as section_with_name,
     pages.subsection_with_name as subsection_with_name,
-    date(navigation.emission_time) as visited_on
+    max(date(navigation.emission_time)) as visited_on,
+    pages.subsection_block_id as subsection_block_id,
+    pages.section_block_id as section_block_id
 from {{ ref("navigation_events") }} navigation
 join
     {{ ref("dim_course_blocks") }} blocks
@@ -27,3 +29,14 @@ join
         and pages.section_number = blocks.section_number
         and pages.subsection_number = blocks.subsection_number
     )
+group by
+    org,
+    course_key,
+    block_id,
+    course_order,
+    actor_id,
+    page_count,
+    section_with_name,
+    subsection_with_name,
+    subsection_block_id,
+    section_block_id

models/navigation/fact_pageview_engagement.sql

@@ -3,30 +3,33 @@ with
         select
             org,
             course_key,
-            course_order,
             actor_id,
             'section' as section_content_level,
             'subsection' as subsection_content_level,
             page_count,
             count(block_id) as pages_visited,
             section_with_name,
-            subsection_with_name
+            subsection_with_name,
+            subsection_block_id,
+            section_block_id
         from {{ ref("fact_navigation_completion") }}
         group by
             org,
             course_key,
-            course_order,
             actor_id,
             page_count,
             section_with_name,
-            subsection_with_name
+            subsection_with_name,
+            subsection_block_id,
+            section_block_id
     ),
     pageview_engagement as (
         select
             org,
             course_key,
-            course_order,
             actor_id,
+            sum(pages_visited) as pages_visited,
+            sum(page_count) as page_count,
             case
                 when pages_visited = 0
                 then 'No pages viewed yet'
@@ -36,48 +39,44 @@ with
             end as section_subsection_page_engagement,
             section_with_name,
             section_subsection_name,
-            content_level
+            content_level,
+            block_id
         from pageview_section_subsection ARRAY
         join
             arrayConcat(
                 [subsection_with_name], [section_with_name]
             ) as section_subsection_name,
             arrayConcat(
                 [subsection_content_level], [section_content_level]
-            ) as content_level
+            ) as content_level,
+            arrayConcat([subsection_block_id], [section_block_id]) as block_id
         group by
             org,
             course_key,
-            course_order,
             actor_id,
-            section_subsection_page_engagement,
             section_subsection_name,
             section_with_name,
-            content_level
-    ),
-    final_results as (
-        select
-            pageview_engagement.org as org,
-            pageview_engagement.course_key as course_key,
-            pageview_engagement.section_subsection_name as section_subsection_name,
-            pageview_engagement.content_level as content_level,
-            pageview_engagement.actor_id as actor_id,
-            pageview_engagement.section_subsection_page_engagement
-            as section_subsection_page_engagement,
-            pageview_engagement.section_with_name as section_with_name,
-            pageview_engagement.course_order as course_order,
-            users.username as username,
-            users.name as name,
-            users.email as email
-        from pageview_engagement
-        left outer join
-            {{ ref("dim_user_pii") }} users
-            on (
-                pageview_engagement.actor_id like 'mailto:%'
-                and SUBSTRING(pageview_engagement.actor_id, 8) = users.email
-            )
-            or pageview_engagement.actor_id = toString(users.external_user_id)
-        where section_subsection_name <> ''
+            content_level,
+            block_id
+    )
+select
+    pageview_engagement.org as org,
+    pageview_engagement.course_key as course_key,
+    pageview_engagement.section_subsection_name as section_subsection_name,
+    pageview_engagement.content_level as content_level,
+    pageview_engagement.actor_id as actor_id,
+    pageview_engagement.section_subsection_page_engagement
+    as section_subsection_page_engagement,
+    pageview_engagement.section_with_name as section_with_name,
+    users.username as username,
+    users.name as name,
+    users.email as email
+from pageview_engagement
+left outer join
+    {{ ref("dim_user_pii") }} users
+    on (
+        pageview_engagement.actor_id like 'mailto:%'
+        and SUBSTRING(pageview_engagement.actor_id, 8) = users.email
     )
-select *
-from final_results
+    or pageview_engagement.actor_id = toString(users.external_user_id)
+where section_subsection_name <> ''

models/navigation/schema.yml

@@ -67,6 +67,12 @@ models:
       - name: visited_on
         data_type: Date
         description: '{{ doc("emission_time") }}'
+      - name: subsection_block_id
+        data_type: String
+        description: '{{ doc("subsection_block_id") }}'
+      - name: section_block_id
+        data_type: String
+        description: '{{ doc("section_block_id") }}'
 
   - name: fact_pageview_engagement
     description: "A view for analyzing the number of page views per learner per section and subsection"

models/problems/dim_problem_coursewide_avg.sql

@@ -0,0 +1,79 @@
+{{
+    config(
+        materialized="materialized_view",
+        engine=get_engine("ReplacingMergeTree()"),
+        order_by="(org, course_key, problem_id, actor_id)",
+        primary_key="(org, course_key, problem_id, actor_id)",
+    )
+}}
+
+with
+    last_response as (
+        select
+            org,
+            course_key,
+            problem_link,
+            problem_id,
+            actor_id,
+            course_order,
+            graded,
+            case when success then attempts else 0 end as success_attempt,
+            case when not success then attempts else 0 end as incorrect_attempt
+        from {{ ref("dim_learner_last_response") }}
+    ),
+    coursewide_attempts as (
+        select
+            org,
+            course_key,
+            problem_id,
+            avg(success_attempt) as avg_correct_attempts,
+            avg(incorrect_attempt) as avg_incorrect_attempts,
+            cast(countIf(success_attempt > 0) as Float32)
+            / count(*) as coursewide_percent_correct
+        from last_response
+        group by org, course_key, problem_id
+    )
+select
+    last_response.org as org,
+    last_response.course_key as course_key,
+    last_response.course_order as course_order,
+    last_response.problem_link as problem_link,
+    last_response.graded as graded,
+    last_response.actor_id as actor_id,
+    last_response.problem_id as problem_id,  -- used for primary key
+    cast(
+        coursewide_attempts.avg_correct_attempts as Float32
+    ) as avg_correct_attempts_coursewide,
+    cast(
+        coursewide_attempts.avg_incorrect_attempts as Float32
+    ) as avg_incorrect_attempts_coursewide,
+    cast(
+        coursewide_attempts.coursewide_percent_correct as Float32
+    ) as coursewide_percent_correct,
+    last_response.success_attempt as correct_attempts_by_learner,
+    last_response.incorrect_attempt as incorrect_attempts_by_learner,
+    cast(
+        CountIf(success_attempt > 0) / count(1) as Float32
+    ) as selected_learner_percent_correct,
+    cast(
+        CountIf(incorrect_attempt > 0) / count(1) as Float32
+    ) as selected_learner_percent_incorrect
+from last_response
+join
+    coursewide_attempts
+    on last_response.org = coursewide_attempts.org
+    and last_response.course_key = coursewide_attempts.course_key
+    and last_response.problem_id = coursewide_attempts.problem_id
+group by
+    org,
+    course_key,
+    course_order,
+    problem_link,
+    problem_id,
+    graded,
+    actor_id,
+    avg_correct_attempts_coursewide,
+    avg_incorrect_attempts_coursewide,
+    coursewide_percent_correct,
+    correct_attempts_by_learner,
+    incorrect_attempts_by_learner

models/problems/dim_problem_responses.sql

@@ -2,7 +2,7 @@
     config(
         materialized="materialized_view",
         engine=get_engine("ReplacingMergeTree()"),
-        primary_key="(course_key, block_id_short)",
+        primary_key="(course_key, block_id_short, response)",
         order_by="(course_key, block_id_short, response)",
     )
 }}

models/problems/dim_problem_results.sql

@@ -2,7 +2,7 @@
     config(
         materialized="materialized_view",
         engine=get_engine("ReplacingMergeTree()"),
-        primary_key="(course_key, block_id_short)",
+        primary_key="(course_key, block_id_short, actor_id)",
         order_by="(course_key, block_id_short, actor_id)",
     )
 }}

models/problems/dim_subsection_performance.sql

@@ -0,0 +1,51 @@
+{{
+    config(
+        materialized="materialized_view",
+        engine=get_engine("ReplacingMergeTree()"),
+        primary_key="(org, course_key, block_id, score_range)",
+        order_by="(org, course_key, block_id, score_range)",
+    )
+}}
+
+with
+    avg_actor as (
+        select org, course_key, block_id_short, avg(scaled_score) as avg_score, actor_id
+        from {{ ref("dim_subsection_problem_results") }}
+        group by org, course_key, block_id_short, actor_id
+    ),
+    avg_total as (
+        select org, course_key, block_id_short, avg(scaled_score) as total_avg
+        from {{ ref("dim_subsection_problem_results") }}
+        group by org, course_key, block_id_short
+    ),
+    score_ranges as (
+        select
+            org,
+            course_key,
+            block_id_short,
+            case
+                when round(avg_score * 100, 2) > 90
+                then '>90%'
+                when round(avg_score * 100, 2) > 70
+                then '71-90%'
+                when round(avg_score * 100, 2) > 50
+                then '51-70%'
+                when round(avg_score * 100, 2) > 30
+                then '31-50%'
+                when round(avg_score * 100, 2) > 0
+                then '1-30%'
+                else '0%'
+            end as score_range,
+            count(1) as score_range_count
+        from avg_actor
+        group by org, course_key, block_id_short, score_range
+    )
+select
+    score_ranges.org as org,
+    score_ranges.course_key as course_key,
+    score_ranges.block_id_short as block_id,
+    cast(round(avg_total.total_avg, 2) as Float32) as total_avg,
+    score_ranges.score_range as score_range,
+    score_ranges.score_range_count as score_range_count
+from score_ranges
+join avg_total using (org, course_key, block_id_short)

models/problems/dim_subsection_problem_results.sql

@@ -2,8 +2,8 @@
     config(
         materialized="materialized_view",
         engine=get_engine("ReplacingMergeTree()"),
-        primary_key="(course_key, block_id_short)",
-        order_by="(course_key, block_id_short, actor_id)",
+        primary_key="(course_key, block_id_short, actor_id, problem_id)",
+        order_by="(course_key, block_id_short, actor_id, problem_id)",
     )
 }}
 
@@ -56,6 +56,7 @@ select
     org,
     course_key,
     block_id_short,
+    problem_id,
     problem_number,
     problem_name_location,
     actor_id,

models/problems/fact_problem_engagement.sql

@@ -3,7 +3,6 @@ with
         select
             problems.org as org,
             problems.course_key as course_key,
-            items.subsection_course_order as course_order,
             problems.actor_id as actor_id,
             items.item_count as item_count,
             problems.problem_id as problem_id,
@@ -32,7 +31,6 @@ with
         select
             org,
             course_key,
-            course_order,
             actor_id,
             'section' as section_content_level,
             'subsection' as subsection_content_level,
@@ -46,7 +44,6 @@ with
         group by
             org,
             course_key,
-            course_order,
             actor_id,
             item_count,
             section_block_id,