49 replies to this topic

[vandalhooch]

RussianWolf, on 08 July 2017 - 12:33 PM, said:

Won't bother checking your math. It's been better than 20 years since I took my one and only Stats course, but I did ace it at the time. Just don't have much need for it in my job.

Who cares if you "aced" your class? Either the analysis supports your claim or it doesn't.

Quote

So you have a decreasing trend that you can only explain roughly half the variance with a decrease in population. So the 20% overall drop indicates that 10% is a result of a decrease of population.

So much for "acing" your stats class. You don't seem to know what variance is.

Take note that player numbers did not decrease every month. Some months they went up. Over half of the variance seen in the data set is due to "random" fluctuations.

Quote

Since anything over 5% is generally considered a significant percentage. I think you may have confirmed what I said.

Statistical significance with correlation coefficients are problematic as it varies depending on what you are trying to model. In general though, -0.69 is just barely at the edge of significance for a sample size of 11 points.

Edit: Since when is 5% considered a significant percentage? Are you confusing typical alpha values for percentage drop between the beginning and end of the data series? Those are two completely different things.

Quote

That MWO has a significant drop in population over either scale. (by the way, I "cherry picked" the first numbers since they were the most recent 6 months and all the data for 2017, You know we accountants usually confine things to a given year, right?)

In other words, you introduced an artificial factor to the analysis because . . . ?

What does a particular calendar year have to do with the playerbase for an online video game? There are no tax code implications for any of the data.

Edited by vandalhooch, 08 July 2017 - 12:49 PM.

[RussianWolf]

vandalhooch, on 08 July 2017 - 12:46 PM, said:

Who cares if you "aced" your class? Either the analysis supports your claim or it doesn't.



So much for "acing" your stats class. You don't seem to know what variance is.

Take note that player numbers did not decrease every month. Some months they went up. Over half of the variance seen in the data set is due to "random" fluctuations.



Statistical significance with correlation coefficients are problematic as it varies depending on what you are trying to model. In general though, -0.69 is just barely at the edge of significance for a sample size of 11 points.

Edit: Since when is 5% considered a significant percentage? Are you confusing typical alpha values for percentage drop between the beginning and end of the data series? Those are two completely different things.



In other words, you introduced an artificial factor to the analysis because . . . ?

What does a particular calendar year have to do with the playerbase for an online video game? There are no tax code implications for any of the data.

Accounting isn't about taxes.

[vandalhooch]

RussianWolf, on 08 July 2017 - 01:19 PM, said:

Accounting isn't about taxes.

And statistics is harder than "I aced an intro class 20 years ago and have done nothing since."

[RussianWolf]

vandalhooch, on 08 July 2017 - 01:23 PM, said:

And statistics is harder than "I aced an intro class 20 years ago and have done nothing since."

I didn't bring statistics into it at all.

[vandalhooch]

RussianWolf, on 08 July 2017 - 01:40 PM, said:

I didn't bring statistics into it at all.

Your first post: "Just looking at season 6 through 12... almost a 25% drop in 6 months. That's.... not..... good."

I especially like the part where you were looking at all the seasons of data and just chose to start with season 6.

You looked at a series of data points, did a calculation and drew a conclusion. That's statistics. Not my fault you had no idea what you were doing.

[RussianWolf]

vandalhooch, on 08 July 2017 - 01:46 PM, said:

Your first post: "Just looking at season 6 through 12... almost a 25% drop in 6 months. That's.... not..... good."

I especially like the part where you were looking at all the seasons of data and just chose to start with season 6.

You looked at a series of data points, did a calculation and drew a conclusion. That's statistics. Not my fault you had no idea what you were doing.

Yes, I chose to start with January.

Okay, by the basic definition I made a statistical analysis that any grade schooler could do.

You smart.

[vandalhooch]

RussianWolf, on 08 July 2017 - 01:59 PM, said:

Yes, I chose to start with January.

Except what you quoted did not have the word January in it. It just said Season 6 and happened to have the highest player population for any season in the series. Some coincidence.

Quote

Okay, by the basic definition I made a statistical analysis that any grade schooler could do.

You smart.

[RussianWolf]

vandalhooch, on 08 July 2017 - 02:08 PM, said:

Except what you quoted did not have the word January in it. It just said Season 6 and happened to have the highest player population for any season in the series. Some coincidence.

yes and I analyzed the data points, did a rudimentary calculation and was able to conclude that season 6 was in fact January. I guess that was statistics also.

Really, use common sense much?

Edited by RussianWolf, 08 July 2017 - 02:35 PM.

[vandalhooch]

RussianWolf, on 08 July 2017 - 02:34 PM, said:

yes and I analyzed the data points, did a rudimentary calculation and was able to conclude that season 6 was in fact January. I guess that was statistics also.

Really, use common sense much?

Cherry picked your data.

[nehebkau]

VirtualRiot, on 06 July 2017 - 04:05 PM, said:

Do you guys think the global leader-boards for matches played could be used as an indicator of the number of players who play?

I'm not sure when each "season" started and ended, or whether certain players are or are not on the board, but going back to season 4 these are the number of people on the board;

S13 - 17204 <------- Current Season
S12 - 31552
S11 - 33727
S10 - 33105
S9 - 33672
S8 - 34663
S7 - 38102
S6 - 41050
S5 - 55205
S4 - 56774

Excluding the current season as it is not yet over, that puts the calculation using a simple linear equation at ~ 24290 players listed at the end of the current season, with 21196 by season 14.

I understand that the regression for a linear equation with this data set is probably only around 0.8, so its probably more like 29148 for season 13 and 25435 for season 14 but there is a decrease over time for sure.

I also seem to be running into more and more potatoes in T1 games as time goes on...

Who knows, maybe this data is worthless due to a multitude of potential reasons, I just thought it was interesting.

Use your steam stats to count unique players. A survey earlier showed a 60%/40% split steam vs regular client so, you can get a good indication of how many people have played.