Highlights from Day 2 of the 2014 HEDW conference.
At breakfast I met someone from Ithaca College who asked me if my school was considering a hosted cloud platform for our data warehouse. We are not. She is. The small BI unit at Ithaca is eager to lighten their burden of routine maintenance tasks. Later that day I watched this person repeatedly pose her question to anyone around her during meals and breaks. I might try her systematic networking method some day when faced with a decision to which there is no clear answer.
I attended “Effective practices for data governance & lessons learned” presented by University of Notre Dame Information Services staff. Governance, a key theme of their presentation, was a trending theme at the conference. In this presentation I saw the most engaging illustration of why the same data question can have 2 answers.
Every data request has these three components – time, element, and context. Time and element are usually explicitly expressed by the data requester. Context – not so much. “How many faculty do we have in current fiscal year” answers time and element. Context is missing, and might be “…who’s primary role is faculty” (n=53), which eliminates the people with a primary role = staff but who teach one class (n=5). Thus, there could be 58 faculty or 53, depending on whether the context of primary role was important. This presentation motivates me to encourage data users at Rochester to make their assumed context explicit – in conversation, metadata, and email.
We had a rainy walk to dinner at Pod – a great pan-Asian restaurant with an all-white George Jetson decor. Within our group were colleagues from George Washington and Stanford. Both schools had signed with Collibra in the past 60 days – the first two US schools to sign. It will be interesting to see how they utilize Collibra.
The University of Pennsylvania hosted the 11th annual conference for members of the Higher Education Data Warehouse organization. Visit the conference website.
Highlights of Day 1:
From the keynote speaker:
- Balance agility with data management. Produce reports and dashboards quickly, but do not cut corners with the quality and repeatability provided by data management.
- Best Practice winners always execute well on their ability to govern and align with business needs.
- Know the business needs. Know the business needs. Know the business needs. Know…
Session 1 – Visualizing Data Using Mapping Tools
Graph mapping tools:
Sources for mapping design inspiration:
Clearly and Simply – a blog not currently active but useful historical design examples.
Session 2: Journey in Development of a Hybrid BI reporting system. This session summarized Oregon State’s efforts to develop a data driven university.
- Emphasize university-wide tools, report templates, and data.
- Create an open access environment where data is easily available to many staff. They made a conscious decision to implicitly trust staff and offered a process to address data abuse when it occurred. One creative solution: Published an access audit on individual staff records. Staff can see who accessed their record, and when access occurred. This public audit discourages random data access by curious staff.
- Accommodate schools within the Oregon system who need exceptions.
- Publish pathways to standard data sets. Publish a score card for each data set indicating quality and and suggested use cases.
Session 3. Analytics to support performance based funding.
New Mexico State University has a low retention – something like mid-40s%. This puts performance based funding at risk. Dashboards or portals tracking retention, completion and progression are trending in many schools as funding sources shift to performance-based. This program was about the details of New Mexico’s excellent analytics on this issue. My take-away is that this is a trending topic.
Institutional Research staff from Georgia Tech did a great job show-casing their self-service dashboard for report delivery (available publicly) at the Data Cookbook User Group meeting.
Their IR team offers 6 self service reports and a term definition search widget. Design elements are clean and easy to understand:
Clicking the Enrollment Report square generates a report on the fly, with user-controlled filters by date and full/part time.
Clicking the small “?” icon near the report title – Enrollment By Major… calls the API to display the report specification from Data Cookbook.
Hovering the “?” near each column label displays the definition for that term via another Cookbook API. In this screen shot, the “Level” column is highlighted.
This is the kind of integration between reporting and metadata that we are striving for at Rochester.