 |
Page History
Goals
| Goal | Accomplished? | |
|---|---|---|
| 1 | Produce version of system for demonstration. | |
| 2 | Define team GitHub process and educate the team. | Pushed to Sprint 3 |
| 3 | Load data for the majority (80%) of nodes for one dog. | Changed to 50% on 2/12/19 |
| 4 | Validate data loading by basic searching/visualizing of the data. | |
| 5 | Integrate Pentaho into the transformation process (timeboxed as an experiment for this sprint). | Yes - Philip |
Retrospective Results
| Author | Category | Card Title |
|---|---|---|
| Philip M |
| What Worked | The idea of manually defined, static but real/realistic seed data being used to generate tsv files with which to pre-configure structure around studies, study arms, and corresponding cohorts. This has given us a robust, scalable mechanism via which to quickly create a realistic study landscape that we can then import subjects and their clinical data into. We should extend this approach in Sprint 3 and mechanize the associated generation of tsv files. | |
| Philip M | What Worked | Ye being able to iterate the data model in real time during our initial data loads allowed us to immediately address gaps in properties associated with target nodes, and/or fully align field nomenclature before trying to load a given set of data. The pay off from Mark's extra effort on the Make Model tool was huge. |
| Matthew B | What Worked | Team continued great work while PM out on vacation. |
| Amit M | What Worked | Separation of file generation and load testing by different individuals. |
| Philip M | What Worked | Collaboration and coordination around data loading was excellent. |
| Amit M | What Worked | Planning out the next set of events and identifying roles and responsibilities |
| Amit M | What Didn't Work | Mark being Mark! |
| Philip M | What Didn't Work |
| Use of manually-contrived values for Submitter ID in all nodes downstream of Cycle, rather than IDs derived from actual data values using Pentaho-based transformations. They worked in terms of allowing data imports in the extreme short term, but simply don't represent a realistic, scalable solution. | ||
| Philip M | Suggestions for Improvement | As a team, we need to maintain better focus in our meetings. |
Bugs
Future
- Take model and impose initial data constraints (add properties).
- Conduct dry-run of Demo.
- Conduct Demo.
Tickets
| Jira | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
|
