1341717 - Write "Common Analysis Gotchas"

Status: RESOLVED FIXED

(Data Platform and Tools :: Documentation and Knowledge Repo (RTMO), defect, P1)

Description

[Ryan Harter [:harter]]

This document will highlight some common data issues encountered when analyzing telemetry data. For example:
* User vs Session weighting
* Profiles vs Users
* Opt-in vs Opt-out
* Dates and their caveats

These topics should not be

Comment 1

[Ryan Harter [:harter]]

{hit submit too soon}

These topics do not need to be covered extensively. The goal is to make the reader aware of these problems so they know to be cautious when building analyses sensitive to these problems. 

We should have extensive documentation for these issues either in RTMO or in the "Datasets and Data Collection" section.

Comment 2

[Chris H-C :chutten]

Looking to have this done for Q2.

Comment 3

[Mark Reid [:mreid]]

Will this cover the date-related topics in Bug 1301820 too?

Comment 4

[Chris H-C :chutten]

I wasn't planning on it... but upon reflection, it seems like a good subsection to cover under this broader heading. "Time is Hard, and pings don't make it easy"

Could cover the different types of delay, the different date formats. I suppose the perils of having high-resolution client timestamps aren't really _analysis_ gotchas, but maybe I could drive-by the problem.

Comment 5

[Ryan Harter [:harter]]

Moving to P2 since we're hoping to have this in Q2.

Comment 6

[Chris H-C :chutten]

Merged: https://github.com/mozilla/firefox-data-docs/pull/13