Reducing the size and removing the unused measures in the Tabular model at IFCJ:
Background:
Some of the Tabular models become too big, causing refreshes to fail and the report’s speed became an issue (some visuals run more than 3 minutes).
Solution:
By utilizing Dax Studio and Vertipaq Analyzer tools, I was able to pinpoint bottlenecks and remove some of the unused large-size columns. For other columns, I substituted key formatted as text with numeric values. Then, I re-wrote some of the DAX codes to improve the speed of the slowest dimensions. Lastly, I worked with the business team members to keep only the data that they need (3 years only vs 10+ years). After optimization, the largest model went down in size by 40%, and most of the visuals could load in less than 20 seconds. At a later point, I revisited the same model and removed ~40% of its measures. Please see the Tabular Models tab for a process review that allowed the removal of unused measures.
BIME Dashboard DeVry:
Background:
DeVry University utilized static excel reports that contained downstream success metrics (e.g. applications, enrollments, memberships). These reports did not provide a comprehensive picture of marketing activities in one place. Also, major channels (paid search, display, landing pages, and social) were managed by different agencies.
Solution:
Senior leadership acquired a proprietary dashboard and I led both technical and business aspects of its development. The new dashboard covered a big picture metrics on the first page (Unemployment, spent, competitor’s spent), followed by internal success metrics. All metrics were compared against similar data for the previous recruitment cycle. As a result of my work on DeVry’s dashboard, I was awarded the Ron Taylor Spot Award.
