529 réponses à ce sujet

[GuyBlade]

DHR107 wrote...

Reviving wrote...

I'm also guessing that you would be able to make the best estimate of the UR drop rate, given that you know the number of gold and UR cards that a person has unlocked. Of course it needs to be before they start unlocking lvl IV ammo, so you'd have to only pick manifests where rares were not maxed.

This is wrong. I have Level 4 Ammo being unlocked even after the new gold weapons/characters from the Reckoning DLC. Mk 4 Ammo can be gained at any time, regardless of whether or not a manifest has been maxed. 

On topic though: These are amazing stats... Actually makes me weep when I see how many more crates I need to get everyhting I need to get hold of!

Based on what I've seen Level 4 equipment begins dropping when you've maxed one of the categories of unlockables: characters, weapons, gear, weapon mods. The RNG store seems to roll a category then pick randomly from within the category. If you already have everything in the category, then it spits out a level 4 ammo.

For instance, I maxed my rares before Reckoning and was thus getting L4 ammo. Last night I finished up the two new peices of gear (though I still have some in each of the other categories to unlock) and I've already seen some L4 ammo.

I'm also of the opinion that the new rare weapon mods will only show up in slot number three of a PSP, but that may be me drawing conslusions based on a small sample set.

[GodLovesATrier]

GuyBlade wrote...

Unfortunately, I don't store the data in an easily queriable way because I was pretty lazy. I did throw together a quick script to answer the question, anyway:

I have about 1800 people who have their badge information currently cached. I've wiped cache information a few times in the past, mainly when I've started collection new data. Based on what  I have, here are the aggregate stats:

Rares / hour    0.022781022639514
Urares / hour:  0.008764278675299
Avg N7 lvl: 1483.6

Now, the thing to remember here is that the most avid players are probablly overrepresented here and once they've maxed their manifest, they start driving down the x/hour stats since the max rares/urares is fixed but play time can continue essentially forever.

Thanks for checking, hope this isn't generating work or too much distraction for you. Good point about the stats being skewed by the most dedicated players. You can see that by the avg N7 level, I'm pretty sure that is top 1-2 percentile of N7. I don't think it will be possible to calculate R/H for the reasons you mention (and lvl IV ammo and character cards before that) and probably wouldn't be that interesting a number for most in the community anyway as the results are fairly predictable (you know how many golds drop in a pack etc). The UR/H number looks a bit off though, is it really 0.009?

[GuyBlade]

Reviving wrote...

GuyBlade wrote...

Unfortunately, I don't store the data in an easily queriable way because I was pretty lazy. I did throw together a quick script to answer the question, anyway:

I have about 1800 people who have their badge information currently cached. I've wiped cache information a few times in the past, mainly when I've started collection new data. Based on what  I have, here are the aggregate stats:

Rares / hour    0.022781022639514
Urares / hour:  0.008764278675299
Avg N7 lvl: 1483.6

Now, the thing to remember here is that the most avid players are probablly overrepresented here and once they've maxed their manifest, they start driving down the x/hour stats since the max rares/urares is fixed but play time can continue essentially forever.

Thanks for checking, hope this isn't generating work or too much distraction for you. Good point about the stats being skewed by the most dedicated players. You can see that by the avg N7 level, I'm pretty sure that is top 1-2 percentile of N7. I don't think it will be possible to calculate R/H for the reasons you mention (and lvl IV ammo and character cards before that) and probably wouldn't be that interesting a number for most in the community anyway as the results are fairly predictable (you know how many golds drop in a pack etc). The UR/H number looks a bit off though, is it really 0.009?

So, I did a bit more looking. After excluding invalid data and properly counting everything (I was previously only counting rare weapons), I reran the numbers and got something that is probably a bit truthier:


Rares/hour:      1.1120374227339
Urares / hour:  0.0082019393611387
Avg N7 lvl:         1560.00061

I'm still only counting UR weapons in the ultra rares per hour above. This is mostly because I think the UR characters have a (substantially) higher drop rate than the UR weapons. The rares per hour went up substantially, but the UR number still seems to be the same, even slightly lower.

Modifié par GuyBlade, 04 mars 2013 - 10:47 .

[GodLovesATrier]

You are definitely right about the UR characters vs weapons. I believe that one of the devs confirmed that they have a different drop rate entirely.

Is there a decimal point error in the UR/H number though - 0.008 implies one UR per 120 hours or so? My UR/H is 0.13 which I suspect is average to low side.

You have even better data than the collaborative spreadsheet for estimating the UR drop rate if such a thing is of interest and indeed if it's even been constant over time.

[Logos77]

 Gracias!

[GuyBlade]

Reviving wrote...

You are definitely right about the UR characters vs weapons. I believe that one of the devs confirmed that they have a different drop rate entirely.

Is there a decimal point error in the UR/H number though - 0.008 implies one UR per 120 hours or so? My UR/H is 0.13 which I suspect is average to low side.

You have even better data than the collaborative spreadsheet for estimating the UR drop rate if such a thing is of interest and indeed if it's even been constant over time.

Oh, how I love PHP. It doesn't warn about calling count() on a non-array. Turns out that I was only counting the 0th item in the array (either the Saber or the Spitfire depending on how old the cached data is). Even truthier data:


Urares / hour: 0.13050272671506

[K9lArthur]

Thanks blade, it makes more sense now.

[Samerandomscreennameidontcareabout]

Yeah, still a big thanks from a long time user of this thingie. - and also thanks for the quick update of the Retaliation manifest.

[ED209-XL]

Neat!

[GodLovesATrier]

GuyBlade wrote...

Oh, how I love PHP. It doesn't warn about calling count() on a non-array. Turns out that I was only counting the 0th item in the array (either the Saber or the Spitfire depending on how old the cached data is). Even truthier data:

Urares / hour: 0.13050272671506

PHP = Rapid application development at it's finest!

Thanks for this, so I am absolutely average at 0.13 myself. Paranoia about low drop rates was just that.

On a statistical note for this sort of data if you are worried about those that have already maxed their manifests or those that have not unlocked any UR's polluting the data then median might be a better measure than mean. I think you have a large enough sample that there won't be much difference though.

[GuyBlade]

Reviving wrote...

GuyBlade wrote...

Oh, how I love PHP. It doesn't warn about calling count() on a non-array. Turns out that I was only counting the 0th item in the array (either the Saber or the Spitfire depending on how old the cached data is). Even truthier data:

Urares / hour: 0.13050272671506

PHP = Rapid application development at it's finest!

Thanks for this, so I am absolutely average at 0.13 myself. Paranoia about low drop rates was just that.

On a statistical note for this sort of data if you are worried about those that have already maxed their manifests or those that have not unlocked any UR's polluting the data then median might be a better measure than mean. I think you have a large enough sample that there won't be much difference though.

Medians:

Rares/hour median:    1.2544346524171
URares/hour median: 0.12161014629328

Remember that my sample size isn't huge (just shy of 1700 people) and isn't necessarily randomly distributed.

Modifié par GuyBlade, 05 mars 2013 - 04:50 .

[GuyBlade]

And, as long as we're posting statistics, here are the standard deviations that go with the means seen earlier:

Rares/hour stdev: 0.7471438257554
URares/hour stdev: 0.09198475477756

[AizenSosuke1984]

Thanks for bumping this thread Blade, this is amazing. Does anyone know how I can add my badge to my signature?

[Uchimura]

On the left side go to profile > forum settings. Put the url inside image tags like this [ img ] url [ /img ] without the spaces.

Modifié par Uchimura, 05 mars 2013 - 05:35 .

[AizenSosuke1984]

Uchimura wrote...

On the left side go to profile > forum settings. Put the url inside image tags like this [ img ] url [ /img ] without the spaces.

Thanks

[GuyBlade]

I updated the badges to reflect the fact that there are apparently no longer promotional weapons. It may take your badge a little while to update, but if you see any lingering problems, let me know.

[Volkai7]

Ohhhh so that's what those are! I've been wondering about those for months!

By the way, both the OP on this thread and your Badges page fail to explain what Purple means for the Readiness badge. Presumably it means you have 255, but that's never actually explicitly stated.

Modifié par Volkai7, 13 mars 2013 - 01:25 .

[GuyBlade]

Volkai7 wrote...

Ohhhh so that's what those are! I've been wondering about those for months!

By the way, both the OP on this thread and your Badges page fail to explain what Purple means for the Readiness badge. Presumably it means you have 255, but that's never actually explicitly stated.

Actually, it means that you have > 250. I added it when I hit 255 in three non-Medigel categories so that I could tell when it was time to buy a jumbo equipment pack.

[GuyBlade]

Also, I updated the badges page to explain it.

[Volkai7]

Thanks, that makes a lot of sense!

[121210Olivia]

Awesome work Guy, been wearing this since it came out ;-)

[darkpassenger2342]

121210Olivia wrote...

Awesome work Guy, been wearing this since it came out ;-)

[Gunboat Diplomat]

Thanks!

[MASSKETEER]

Very very nice job!!!! Thank you Guyblade

[N172]

I found a bug in the Promotion Distribution badge:
The Vanguard Portion is too large.

For example, I currently got:
82 times Adept
78 times Infiltrator
75 times Soldier
73 times Engineer and Sentinel
71 times Vanguard

But on the badge the Vanguard portion is the same size as the Adept portion (larger than the others), it is worse on the uber badge, where Vanguard is the largest.
I think it might be some kind of rounding issue.

Aside from that: Nice Job!