Stats stats stats and stats ! NC hardmode version 2

Yeah I realise that they show the breakdown for each category of weapon. But it still doesn't show me how they derived the KPU for each quarter on each weapon in each row. The total sheet is just the sum of these KPU results and then averaged. The fact that these don't add up correctly in the totals sheet (example I gave as the TR totals) is clear indicator that the breakdown has incorrect data in some or all of the data points. Since these breakdowns only show the result of there KPU calculations and not there workings. I can't confirm their accuracy or find where the issue stems from. It could be a simple issue of incorrect data inputted into these sheets. Or it could be that the underlying formula is incorrect. Or if I wanted to be cynical, it could have been purposely altered after the fact to skew the overall result. Either way, the totals highlight an issue with the data that stems somewhere down the line and has resulted in incorrect data overall. At this point the spreadsheet is just a bunch of meaningless numbers.

The KPU value per raw data line is simply: total kills / uniques

The quartiles per raw data are generated by taking all the BR of the uniques and assigning them into the quartiles. Then the KPU is taken of each quartile. There won't be the same number of uniques in each quartile - but that data is not reported. It's to give an indication of the weapon performance as players become more experienced. The average of the quartile KPU won't be equal to the KPU value. (at least this is how I understand it, could be wrong)

That I understood and from what I have checked is correct. You are also correct in that the average of the quartiles won't be equal to the average KPU value. The problem is they don't state the raw data for each quarter i.e. how many users and kills in each quarter. All they report is the final result for each quarter in the KPU. Therefore I can't check those results. The problem is that these quarter results don't match the overall KPU results. This is most evident in the totals. As I have already stated, it is mathematically impossible for every quarter to be down, yet the total average to be up. You dont need to know the raw data to determine that fact. What this means is that either the quarter KPU results have been inputted wrong or formulated wrong before input. That or someone has altered the total numbers to skew the results, but failed to alter the quarter results. Either one would result in this discrepancy in the data.

You can check this for yourself in the total sheet. While the # of kills, # unique users, and KPU results are a sum of the breakdown sheets. If you apply the same formula to these figure (kills/users) you get completely different result than what is stated in the KPU. NC is 6.7 instead of 6.2, TR is 7.3 instead of 6.4, and VS are 7.0 instead of 6.5. Thats a hell of a big difference and evidence that somewhere down the line the figures are wrong or being formulated incorrectly. The data is simply flawed. And any kind of discussion based on flawed data is irrelevant.

Or, you know, it could just be that someone made a mistake when calculating the totals. All the separate sheets on the different weapon classes have accurate simple averages across days- the figures above are the figures directly from the API.

Not to comment on whether these stats are even valid right now I feel this is appropriate

I like your answer. The point of my "anecdotal evidence" is especially to prove that people who are not able to analyze stats properly should not post about it... Most often the way stats are presented does just make them meaningless. Last point of such "anecdotal evidence" is to prove that the maths most people use on Forumside is just wrong. Doing some additions does not make a point if you don't use the proper formula and proper figures ....

Last, my outfit are some of these NC that fight whatever to try to make a difference in the battle, even if we get steam rolled. So YOU won't tell me what a NC that fights is.

Regards

That's my point : why people raging about NC AI weapons don't use them ?

Yes slugs make them "better" at medium/long range. Better but still a joke : first you have to hit, and even if you hit you need 2 or 3 shots to kill SO... I won't talk about MAX 1vs1 with slugs ^^

I will write in cap letters, maybe this way you would get all of my words.

I USE GRINDERS, I NEVER PLAYED OTHER NC MAX AI WEAPONS, I ****ING WROTE IT NOT IN THE LAST POST YOU QUOTED BUT IN THE PREVIOUS ONE.

So yes, I don't know how to use scatt. If you run out of arguments, please don't use TR-like argument : L2P, you dumb, idiot, you suck.

And please, in order not to be ridiculous, carefuly read the posts you answer.

PS : I won't do the research work to show you in which post of mine you quoted I said this. Just do it yourself.

I would even add that out of these 24h, there are at least 18 hrs in AA Max. So yeah I'm a real noob in MAX AI. I already said it, you quoted it. So please if you run out of arguments, don't use TR-like "L2P, you noob" and ****ing read my posts before answering it. Thanks

The reason is most combat doesn't happen in prime Grinder range (even in Biolabs), so most people want the longest range AI MAX (or at least one with a longer effective range than the Grinder).

Basically now you have the Hacksaw - which is largely pointless as the Grinder is now (since the nerfs) better in that close range role.
And you have the Scattercannon - which is largely pointless as the Mattock or Grinder is always going to be the better option.

So really you have the Grinder with pellets) if you are fighting ~0-8m range or against other MAX or the Mattock with slugs for everything else.

True, but they are the only viable option if you want to extend your range with a NC AI MAX (with Mattocks), bearing in mind TR/VS AI MAX can hold their TTK way out to 25m+.

I'd rather take any other infantry class than my MAX with Mattock+slugs for this role, that's for sure and I guess that the real problem is there.

This is why people are asking (and have been asking for a long time now) for a NC AI MAX LMG type weapon.

Yes there are some calculation errors on the Grand Total sheet, along with an error for the TR assault rifles T1 cycler average uniques. I don't have energy to go through the rest of it. In the end an empire wide KPU is not that useful anyway, as you say it's better to look directly at the weapon classes.

A 2-handed fat LMG would be amazing

That's an interesting idea, I guess they'd need a lot of new MAX animations though. In general it's more Mercy/Blueshift.

Huh, you do realise the general name for the NC MAX weaponry is "Scatters" right? Of course you don't.

It's like calling the piston a semi-auto shotgun. I guess I'm really asking too much out of you. Just leave.

At that point, they might as well call it a HMG or BFG9000.

The totals are the sum of the breakdowns. Which means that the issue is within the breakdown info itself. The totals just highlight the issue.

All the individual sheets provide accurate simple averages across days.

And how do you know they are accurate, when they don't provide the data that they derive the final figure from.