QSDA study plan
QSDA · Give each week a clear focus.
8 weeks · About 6 hours per week
Data Modeling (associations, joins, synthetic keys, circular references, star and snowflake schemas)
Build a three-table star schema from flat source files and verify association direction in the data model viewer
Detect every synthetic key created by duplicate field names and eliminate each one by renaming keys or adding a preceding qualify step
Write a one-line grain statement for each table directly in the script as a comment
Compare row counts before and after every join to prove no fan-out occurred
Your checkpoint
A working app with a documented star schema plus a one-page log listing every synthetic key found, its cause, and the fix applied.
Data Modeling (associations, joins, synthetic keys, circular references, star and snowflake schemas)
Construct a deliberate circular reference between two lookup tables and break it with a link (bridge) table
Break a second circular reference by aliasing the shared key and applying Keep to one side
Flatten a snowflake schema into a star and record the effect on reload time and file size
Compare Join, Keep, ApplyMap, and Mapping Load on the same transformation and note which one preserves the fact-table grain
Your checkpoint
A reference app containing three resolved circular references plus a comparison table showing row counts and reload times for each join technique.
Data Loading and Transformation (load script, ETL patterns, incremental loads, resident and preceding loads)
Rewrite a chain of sequential table loads as nested preceding loads to cut the number of passes over the data
Use a resident load to derive a dimension table from an existing fact table
Add ScriptError handling with a rejected-row log so failures are visible rather than silent
Parameterize source paths and load dates with a single config block of variables
Your checkpoint
A layered script (extract / transform / present) that runs end to end from variables and writes a rejected-row log on a seeded error.
Data Loading and Transformation (load script, ETL patterns, incremental loads, resident and preceding loads)
Implement an incremental load driven by a high-water-mark date persisted in a QVD
Add an insert/update/delete pattern for changed and removed records
Handle late-arriving records and prove that re-running the same reload produces no duplicates
Reconcile the incremental result against a full reload using row counts and totals on three measures
Your checkpoint
An incrementally loaded app in which a same-day re-run produces zero duplicate rows and totals that match a full reload.
Data Quality and Integrity (data cleansing, null handling, granularity, key consistency)
Profile every key field for nulls, blanks, and orphan values and record the counts per field
Apply a consistent null policy (NullAsNull, NullAsValue, or explicit default) and document the choice field by field
Confirm the grain of each fact table and remove duplicate rows at the correct key level
Standardize key formats by trimming, upper-casing, and casting numeric keys consistently
Your checkpoint
A data quality report per table showing row, null, orphan-key, and duplicate counts before and after cleansing.
Performance Optimization (script tuning, data volume reduction, calculation efficiency, QVD usage)
Move the raw extract into a QVD layer and reload the app from QVDs only
Reduce data volume by dropping unused fields and limiting history to a defined date window
Replace repeated inline expressions with variables or master measures and time the calculation impact
Benchmark reload duration and app file size before and after the changes
Your checkpoint
A timed before/after benchmark showing reload time and file size, with the percentage reduction documented.
Security and Governance (section access, data reduction, connection security, governance of published data) & Deployment and Data Architecture Operations (QVD layers, reload scheduling, app and data publishing, environments)
Configure Section Access with a reduction field and validate it with two distinct test logins
Design a three-tier QVD architecture (raw, transformed, presentation) and lay out the folder structure
Script a reload schedule with failure notification and a defined retry path
Document the promotion path for apps and QVDs from development through test to production
Your checkpoint
A published app with Section Access proven by two test logins, plus a one-page diagram of QVD layers, reload schedule, and promotion path.
Data Modeling (associations, joins, synthetic keys, circular references, star and snowflake schemas) & Data Loading and Transformation (load script, ETL patterns, incremental loads, resident and preceding loads) & Data Quality and Integrity (data cleansing, null handling, granularity, key consistency) & Performance Optimization (script tuning, data volume reduction, calculation efficiency, QVD usage) & Security and Governance (section access, data reduction, connection security, governance of published data) & Deployment and Data Architecture Operations (QVD layers, reload scheduling, app and data publishing, environments)
Sit a full-length timed practice exam and score each of the six domains separately
Rebuild from a blank script the two weakest domain artifacts identified by the score card
Consolidate a mistake log that pairs each recurring error type with its corrective pattern
Re-attempt only the missed items and confirm each is now answered correctly with a stated reason
Your checkpoint
A domain-by-domain score card plus a one-page fix list, with every domain scored at or above 70 percent.
Milestones
Week 2: The data model contains no synthetic keys and no circular references, and you can explain the resolution technique used for each one without notes.
Week 4: A full reload and an incremental reload of the same source produce identical row counts and matching totals on at least three measures, with no duplicate-key warnings.
Week 6: The app reloads entirely from QVDs, and you can state the measured reload-time and file-size reduction versus the original script.
Week 8: Every one of the six domains scores at or above 70 percent on a timed full-length practice exam, with Domains 1 and 2 at or above 80 percent.