Incorrect statistics on finds per day of the year?

[+digitalformula]

Hi all,

 

Edit: Question, first.  Are the dates with finds on Geocaching.com showing in my local timezone?  In the PQ I can see they're GMT but I'm in AEST.  Could that be the issue below?

 

I'm working on an app right now and have some inconsistencies between the data I'm getting from the My Finds PQ and what is shown on my statistics page.  For example:

 

- Geocaching.com shows I've found 8 caches on Jan 1st, while the data in the PQ and the data from my app both say I've only found 4 caches on Jan 1st

- Geocaching.com shows I've found 4 caches on Jan 2nd while the data in the PQ and the data from my app both say I've found 8 caches on Jan 2nd

 

If you compare those two things, you'll see that Jan 1st and Jan 2nd are reversed.

 

I'm pretty sure my code is correct as it is showing exactly the same data coming out of the PQ, which I have manually checked on multiple occasions.

 

Can anyone verify this or has anyone else seen it?  I don't want to spend hours trying to figure what's wrong with my app's code if I don't need to.

 

Thanks

Edited February 4, 2018 by digitalformula
Add question

[+MartyBartfast]

I have a suspicion that the PQs report the dates/times in the UTC timezone, so if you convert the dates/times in the PQ from UTC to your own timezone will that correct it?

[+digitalformula]

59 minutes ago, MartyBartfast said:

I have a suspicion that the PQs report the dates/times in the UTC timezone, so if you convert the dates/times in the PQ from UTC to your own timezone will that correct it?

Firstly, you're right.  The PQs contain timestamps showing as GMT (same at UTC, although UTC isn't a timezone, technically speaking).

 

As for converting from GMT to my timezone - nope, that doesn't fix it.  In fact, that shifts almost all dates out by a day. 

 

But, since writing this question I've found some other interesting things.

 

For example, the time associated with a geocaching log seems arbitrary.  I did some testing earlier and logged a cache as found as 21:15.  A PQ generated shortly after showed the cache as logged at 20:00:00 (down to the second).  I did another test shortly before that, logging the cache at 20:30.  The PQ still showed the cache as logged at 20:00.

 

Other logs are showing caches logged at 19:00:00 (also down to the second).  Other caches are showing caches as logged at times like 08:41:25.

 

In fact, the vast majority of logs in my latest 'My Finds' PQ show as logged at either 19:00 or 20:00 - 885 of them, to be precise.  That's out of my 893 finds.  That's also far more than I've actually found at those times so clearly isn't accurate.  E.g. I logged 2 caches as part of a 100-cache series.  They both show as found at the same time on the same day.  So yeah, those times in the PQ clearly aren't accurate.

 

The other weird thing is that I've put my finds per day and the Geocaching.com finds per day data into a spreadsheet for comparison.  Nearly every day throughout the year is showing the same on both, although a collection of seemingly random dates are out by 1.  Never more than 1.  E.g. March 10th shows as having 1 find in my data, but Geocaching.com shows me as having no finds on that day.  Geocaching.com shows a single find on March 9th, though.

 

If nearly every day through the year was incorrect, I'd expect some sort of calculation or date-shifting issue, like an incorrect timezone, but there's no pattern that I can see re the dates that are different.

Edited February 4, 2018 by digitalformula

[+noncentric]

Just looked through a recent My Finds PQ of mine.  All of my Found It logs show a <Groundspeak:date> value of either "T20:00:00Z"  or  "T19:00:00Z".  The one hour difference is probably due to daylight savings.

All but one of my Didn't Find It logs also showed times of either "T20:00:00Z"  or  "T19:00:00Z". The one Didn't Find It log that showed a different timestamp was one that I submitted via the official app.

All of my Needs Maintenance logs showed other timestamps that were not 20:00:00 or 19:00:00.

Since this was a My Finds PQ, it doesn't show all of my DNF/NM logs, just the DNF/NM logs on caches that I later found.

 

Do you see any association between the timestamp formats and the method of your logging (app, website, gsak)?  Most of my finds are logged via the website, and mostly through Field Notes. I'm pretty sure I've logged at least a few caches via the official app, but none of my Found It logs show a specific timestamp.  I'm pretty sure that timestamps are always generalized for Found It logs.

NM logs do not allow date changes, so I suppose that's why the timestamp is locked.

 

On 2/4/2018 at 2:44 AM, digitalformula said:

In fact, the vast majority of logs in my latest 'My Finds' PQ show as logged at either 19:00 or 20:00 - 885 of them, to be precise.  That's out of my 893 finds.

Anything specific about the 8 finds that might indicate the cause?  Logged via a different device?

 

 

[+digitalformula]

1 hour difference - yep, the DST thing seems to be the case.  All the 19:00:00 timestamps seem to be for dates outside of DST.

 

I actually do think the arbitrary/generalised times are for field notes/drafts as that's how I upload and log the vast majority of my finds, via Garmin "geocache_vists.txt".  I need to test this a bit further but I think that's probably the difference.  That said, I'm also fairly certain I've logged more than 8 finds from my phone in 9 years.  I can't be certain until I test but it seems logical that this is the difference.

 

Later tonight, when I get home, I'll fake log a couple of caches from phone, GPS device and website.  I'm pretty I'll find something consistently different (hehe) by testing that. 

 

By the way, Groundspeak responded to my query about this and said they're already aware of some issues with PQs and that the information I provided (same as above) will help them.  Nice of them to respond but it's also nice to know *they* know the PQs aren't perfect.  For unfound PQs the data seems spot on, just not for My Finds PQs.

[+noncentric]

Please do report back on what you found.  Honestly, I wouldn't expect logs to retain timestamps. Some log types allow to select the "Date" for the log, so it might as well use a generic timestamp. Not sure why the logs that don't allow date adjustment retain a specific timestamp.

For example, if I go to a cache page at 8pm on 2/7/2018, click Log Geocache, and select a log date of 1/20/2018, then what timestamp makes sense?  Should it use a timestamp of 8pm, even though I didn't find it at 8pm on 1/20/2018?