Investigate relationship of visual metrics to telemetry metrics.
Categories
(Data Science :: General, task)
Tracking
(Not tracked)
People
(Reporter: ccd, Assigned: ccd)
Details
(Whiteboard: reporter:esmyth)
Brief Description of the request (required):
Visual metrics are the user-centric method for measuring page load. They are being used in our in-house Firefox testing runs (browsertime, Webpagetest). However, they are too computationally demanding to measure in telemetry, as they require video replay of the URI load.
This request is to investigate the relationship between visual metrics and those available in telemetry.
Business purpose for this request (required):
A useful goal would be to utilize the available metrics in telemetry as a proxy for visual metrics. Telemetry probes could then be implemented to measure visual metrics "in-the-wild".
Requested timelines for the request or how this fits into roadmaps or critical decisions (required):
end of Q4
Links to any assets (e.g Start of a PHD, BRD; any document that helps describe the project):
Comparing visual metrics to navigation timings
External and internal Webpagetest instances record many navigation timings in addition to visual metrics. These datasets are 100K-1.5M records. They are available as processed RData files, which populates this dashboard.
This job exports a reduced CSV file of our internal Webpagetest instance.
Name of Data Scientist (If Applicable):
Please note if it is found that not enough information has been given this will delay the triage of this request.
|
Assignee | |
Comment 1•6 years ago
•
|
||
Adding in additional fields in data preparation job of internal Webpagetest (WPT) instance. Figuring out how to explode these fields and then index correctly without manual intervention would be exceedingly useful.
Updated dashboard to support the new fields.
As noted above, these WPT datasets will be very useful for this work.
|
|
Assignee | |
Updated•6 years ago
|
|
|
||
Updated•6 years ago
|
|
||
Comment 2•6 years ago
|
||
Writeup on largest contentful paint in which the correlation to SpeedIndex is calculated (0.83 on mobile and 0.87 on desktop)
https://calendar.perfplanet.com/2019/developing-the-largest-contentful-paint-metric/
|
|
||
Comment 3•6 years ago
|
||
Work for the DS team is now tracked in Jira. You can search with the Data Science Jira project for the corresponding ticket.
Description
•