Stats report: (a) 400+ day active streak but (b) only "364 caching days over 365 total days". Seems a glitch?

Question

Our primary stats page (for bejs) reads as pasted below.  It simultaneously says that we have cached for 409 days in a row (which is accurate), but in the last 365 days we only had 364 caching days.  Does this reflect a technical glitch, or is there something peculiar about our finds that suggests this.  We have never done just Lab Caches on any day - we have always had regular cache finds on the four days we tested out Lab Caches (and decided that at least for now, they are not for us).  Any info would be helpful!  Thanks.  EDIT:  Adding the following paragraph as additional info:

FWIW, I almost always see the "364 caching days over 365 total days" and have at least ever since I crossed passed 365 on my streak.  (Before I passed 365 I would have had no easy way to know if the "xxx caching days over 365" figure was right or wrong.)  So this problem is not just a fleeting issue.

Last 365 days:	1095 finds in 364 caching days over 365 total days (99.7%)
Last 365 days averages:	3.01 finds per caching day, 3.00/day, 21.06/week, 91.25/month
Most consecutive days with finds:	409, from 2021-01-30 to 2022-03-14 (Current streak)
Most consecutive days without a find:	526, from 2016-01-24 to 2017-07-02
Current consecutive days with a find:	409, from 2021-01-30 to 2022-03-14

Comments

Can you tell me what date you have in the Profile stats header? It says something like "Data from Y-m-d".

---

Sorry, saw your longer response before I saw this question.  As I think you correctly figured out, my "Data from" is:  Data from 2022-03-15.  

As I described in my note posted below to your longer response, I would not expect my 03-15 actual data to show up until my "Data from" reads: 2022-03-16.

Thanks.

---

I think the date actually is inclusive. So when it says the 15th, it's data up until and including that date. However, it doesn't necessarily have to be all finds of that date then. But all finds the date before should be there, assuming they had been logged online.

In our database that date is actually a timestamp, so we know up until a certain second all data has been fetched. We use incremental api methods to fetch logs changed (created, deleted, modified) after that timestamp from HQ.

There will always be some syncing issues since (most) geocachers don't live-log everything.

Answer 1

I have looked in the code, and I think I know what happens.

First I'll explain in short how it works. Your data is updated in Project-GC in several ways. Some of them triggers a date that says "this is when all find data was updated". When you load Profile Stats this date is used as a "max date of finds". This is to prevent half days and such, mostly because your finds doesn't necessary come into Project-GC in the correct order.

For paying members this date should normally not be older than 24 hours, for others it can be older. This date and the handling around it is also related to the data-caching of the rendered Profile Stats.

The "Last 365 days" was however using "current date" when it was calculating your finds for the last 365 days. This is logical, normally. But if your "data from date" says that no finds newer than yesterday should be included, well then you can't have 365/365, since today is always missing.

I have created a patch in our development environment that will use the "data from date" instead of current date. This should hopefully resolve this problem.

I hope the explanation is understandable.

Then some bonus info:

• You and Pleu are never seeing the same render when it's your stats. It's because there are privacy settings allowing home location based data for example. Due to that fact you have one cached version, and other users have another cached version.
• There is also the issue around paying vs none paying members. Since none paying members gets a new render every 7th day, while paying gets one every day, the paying members might see newer data on person X.
• There might be other reasons as well that I can't recall right now.

Comments

Excellent and thorough explanation!  Thanks.  Everything you said seems like a rational explanation of the situation.  

For those in the US (rather, at least for me on the East Coast of the US), my stats page info always updates around 9 or 10pm LOCAL East Coast time, but it always says "data from [the next day]".  I have always assumed that the Project-GC servers are in Europe and the server runs in fact happen "tomorrow" from the perspective of my local time.  So assuming the "364 days out of 365 days" bit of data is always looking at my "tomorrow" then of course I cannot have made any finds for that last day.  

This is a glitch that (as noted) I never paid attention to until after my streak was over 365 days.  But I agree that it makes sense to sometime move to have that data point look at the "data from date" instead of the "current date".

Thanks for a very helpful answer, and for all of the work that all of the Project-GCers do.  I was a computer programmer in the late 1970s/early 80s, so my skills are a bit rusty (and languages retired), but I am learning more modern languages now and may someday be able to try to contribute (even if only to tagging challenges, etc.).  Thanks, bejs

---

Your explanation of the problem helped. It's late here and I am tired, been in front of the screen the whole day. But after reading it 2-3 times I could see the issue immediately when I checked the source code, mostly due to your description. Assuming I have been correct that is, that we will see when the patch goes live (it's not yet).

Servers are located in CET (UTC+1). But all times/dates used on the site is UTC, which is the +-0 timezone (like GMT). I can understand how that can feel a bit off if one is UTC-6 for example.

Answer 2

If you check it now it says 365 caching days over 365 total days and a current streak of 410 days. Since it also has your current streak as 2022-03-15 it seem to have picked up todays cache/s. My guess would be that your stats was generated on the 15th before you had logged any caches for that day, but then they shouldn't have been generated again already so that seems a bit odd to me. They're correct now anyways so that might be the most important thing. :)

Comments

Pleu, thanks for the response.  Very interesting.  Unfortunately, no matter how much I refresh my display or use a different browser and on three different devices (Windows, Mac OS, and Android), my stats page still displays the numbers above.  This is true even after I did a Self Support forcing a read of all my finds, and using Ctrl-F5 on my Windows browser to force a full refresh.  

Yes, a little bit before your post, I had logged two finds today, and it looks like your view of Project-GC can see more up-to-date data. You should see 1484 finds not including Labs.  I wonder if you are on the other side of GMT and/or the international date line from me, and that somehow impacts what I can see?  I am on EDT on the East Coast of the US.   

FWIW, I almost always see the "364 caching days over 365 total days" and have at least ever since I crossed passed 365 on my streak.  (Before I passed 365 I would have had no easy way to know if the "xxx caching days over 365" figure was right or wrong.)  So this problem is not just a fleeting issue.

Any thoughts?  FYI, I plan to paste the above paragraph into my original question just so others see it more quickly.  Thanks, bejs

---

That's odd, my understanding is that the stats should be the same no matter who's looking at them.

These are the numbers I'm seeing when I'm looking at https://project-gc.com/ProfileStats/bejs

Last 365 days:    1096 finds in 365 caching days over 365 total days (100.0%)
Last 365 days averages:    3.00 finds per caching day, 3.00/day, 21.08/week, 91.33/month
Most consecutive days with finds:    410, from 2021-01-30 to 2022-03-15 (Current streak)
Most consecutive days without a find:    526, from 2016-01-24 to 2017-07-02
Current consecutive days with a find:    410, from 2021-01-30 to 2022-03-15

And yes, we're on different sides of the GMT-line (I'm at CET(GMT+1)) and I think the server uses GMT, but even if that would mean we can have different dates and that would mess up the statistics we should still see the same numbers. Wouldn't make sense to generate one page to show you and one page to show me when it's half the resources and space to show us the same page. :)