Final major update for GAEN Analyzer, version 1.0.0

Finally released version 1.0.0

• Added the ability to load and save ENPA raw archive so GAEN Analyzer can be used after MITRE server is shut down
• Added the relative risk chart and explanation (used in UK Nature paper on EN)
• Handle situation where there is no Android data
• Fix a hang problem resulting from a 0/0 computation and from an empty sum
• Added new beaconCount analysis showing how many other EN devices are detectable at busy, typical and unbusy times of day.

Major update: national ENPA opt-in, analysis of exposure characteristics, 14 day metrics,

Lots of updates. Major changes:

• Use national ENPA opt-in rather than regional ENPA opt-in calculated from comparing verified codes ENPA/ENCV data. The people who verify codes are not entirely representative of the entire ENPA population, so this gives a more representative estimate, although it doesn't take into account possible regional variables in opt-in %'s. But looking at a few states, it also seems to reflect what is happening overall in terms of ENX adoption.
• added analysis based on the userRisk and userRiskParameters metrics, that show data such as the percent of exposures that had a sumScore > 120 minutes, and the percent of exposure that occurred at less than 60 dB of attenuation.
• analysis of the 14 day metrics, which provide a higher signal to noise ratio, and also better separates out report type.

v0.4.1-b3

Added code to analyze the UserRisk and UserRiskParameters metrics to understand durations and attenuations of exposures. This allows answering questions such as

• How many more notifications would be generated if we changed our exposure threshold from 15 minutes to 10 minutes
• Of the people now getting a notification, how many of those notifications are 60 minutes or longer?

Note: the iOS / iPadOS versions are available through an unlisted entry in the App Store. Contact Bill Pugh for a link.

Added estimated total number of notifications

Added estimated total number of notifications, and smoothed estimated number of users and ENPA opt-in using median of last 14 days.

Add low noise metrics, updates to documentation/text

Updating the information screens for various metrics
Preliminary inclusion of some of the low noise metrics

Full Changelog: v0.2.12-b1...v0.3.1-b1

Change SAR offset back to 0 days, fix export of graph data

Switched back to not using an offset for notifications when computing secondary attack rates
Fixed export of cvs for graphs of individual charts
Adding missing chart documentation
Trimmed trailing dates with no data.

Adding additional graphs

Added graphs of notification share, % of phones showing notification, excess secondary attack rates
removed likely sum from raw export
Don't calculate SAR for days with less than 10 notifications per 100K
Uppercase region (e.g., turn us-md into US-MD).
In summary screen, show region analyzed, not what is in setup
fix bug in case where an SAR contains no usable data.
reorder charts
update documentation to reflect that can use rolling average other than 7 days

v0.2.10-b1

Added ENPA appendix charts: xsar, users with exposure notifications
Changed how SAR% is charted: one graph per category, showing +/- one standard deviation
Try to make bounds on SAR% more sensible.
Changed the offset for SAR to 7 days
Exclude iOS ENPA data before Feb 24th, 2021
Allow using 28 day rolling averages

Bug fixes for loading older encv composite data, other small changes

Some important bug fixes for loading older encv composite data
Allows specification of the number of days to use for rolling averages
A number of smaller changes, such as prefetching ENPA data so that the charts for the selected period include full rollups

First release on GitHub

First release on GitHub of the macOS version. This version includes a lot of changes to make the app work well on iPadOS, and since the macOS version shares much of that interface, it carries over.

One of the big functional changes is to persist ENCV data, since the ENCV server only retains data for 90 days. The app can only load ENCV composite-stats.csv files you have previously downloaded from the ENCV server.

If you can use the macOS version, I recommend that over the iOS version. I can iterate versions of the macOS version more quickly, and things such as transferring API keys and exporting files is slightly easier on the macOS version. Note that the macOS version requires macOS Monterey (macOS 12).

If you don't have convenient access to a computing running macOS Monterey and want to test the app on an iPhone or iPad, contact [email protected].