We don't know where users come from or go to on TMO, and that's an odd sort of
lapse for a data-based web application.
This is preliminary: just gtags for now.
The update orphaning Spark job is now scheduled to run right after the
longitudinal dataset creation job on Sunday, instead of Wednesday. The
dates need to be adjusted as follows:
Monday: -5
Tuesday: -6
Wednesday: -7
Thursday: -8
Friday: -9
Saturday: -10
Sunday: -11
Note that this is in addition to the 7-day adjustment a few lines below
because of the fact that Airflow schedules these jobs with a one week
delay.
We now schedule the weekly Spark job for the update orphaning dashboard
via Airflow. Airflow uses a new URL for the output files. Airflow also
operates on a one-week delay, so we have to drop the most recent date
as an option from the dropdown menu in the dashboard.
Note: older data files will be copied manually to the new location.
1. Change color palette to color-blind-friendly palette for version
distribution pie chart.
2. Change logic to use “versions behind current” instead of actual
version numbers to select colors.
3. Refactored code to generate d3pie charts to have less duplicate code.
1. Reference the responsible person for the dashboard (with email
address).
2. Reference the Spark job that creates the data for the weekly reports.
3. Make it clear in the summary of the page what’s considered
“up-to-date” for each weekly report in the context of this dashboard.
Chris H-C
Robert Strong
Matt Howell
Stephen A Pohl