Graphing the War (Because numbers suck)

[Chad]

Guys, his graph checks out. By March 19th, our side of the war will be at negative nation strength.

Your understanding of math is astoundingly horrible. And the fact that you try to be sarcastic while obviously knowing nothing about how graphs work and lines work is especially alarming. I am, honestly, afraid for the world if this is our future.

[Beets]

The point is that even using sound math (polynomial regression is rock solid, mind you), the prediction is still inaccurate if taken forward more than 5 days or so. The trendline is generated directly from the data and anyone with a graphing calculator (or pencil and paper if you are especially smart) will come to the same conclusion if trying to find a cubic polynomial. Unlike your friend, I did not pull the prediction out of my ass:

Equilibrium Trendline Cubic Polynomial Equation: y= -891.03x3 + 1E+08x2 - 5E+12x + 6E+16

Competence Trendline Cubic Polynomial Equation: y= -1359.9x3 + 2E+08x2 - 7E+12x + 1E+17

You seriously jammed a cubic polynomial equation into those graphs, which is a function that will measure flat lines as something that only appears before it skyrockets downward, then tried to say that's the most accurate thing you could have chosen. That's totally wrong from a math point and that just makes me sad.

Negative root functions or bust.

[Chad]

You seriously jammed a cubic polynomial equation into those graphs, which is a function that will measure flat lines as something that only appears before it skyrockets downward, then tried to say that's the most accurate thing you could have chosen. That's totally wrong from a math point and that just makes me sad.

Negative root functions or bust.

Yes because a linear model would not work. Ideally, I would want a linear model but when projected across the entirety of the data; both Coalitions will see a net gain for the next infinity. Anything higher than a cubic would produce even steeper prediction results. I suppose I could have used a logarithmic trend, but like the linear model, it would show a net gain in NS for both coalitions when paired with the rest of the data. Because other factors are missing (such as the relationship between the two data sets), that honestly was the best trendline I could have used. Perhaps not for the selected data, but when used for prediction in that small one week window, it is ideal. Anything further and it is rendered too inaccurate for use (something which I have already state). But all that being said, here is the same February only data set with a logarithmic model projected over the course of two months:

 

 

A little more realistic looking, yes. But it remains an unusable model when projected with the data from January.

 

And at any rate, my point remains the same. These are the two best mathematical models we can use to compress the data and predict what will happen. Any deviation from these methods leaves the data open to speculation and will therefore be less accurate.

Edited February 22, 2013 by Isotope

[lamuella]

Your understanding of math is astoundingly horrible. And the fact that you try to be sarcastic while obviously knowing nothing about how graphs work and lines work is especially alarming. I am, honestly, afraid for the world if this is our future.

You just gave a trendline that is literally impossible. Expect to have it mocked.

As to knowing nothing about how graphs work: if you're trying to do polynomial regression on a very limited dataset, why didn't you use a confidence band?

[crypticedge]

I have here some completely unbiased graphs that need to be shared with the rest of the game population. Please review them and understand why goons will win. 

[Beets]

Anything higher than a cubic would produce even steeper prediction results.

A cubic always shows flat lines as a prediction that it's going to skyrocket downward.
Going higher would just produce hilariously upward/downward skyrockets depending onwhether it was odd or even. That's hardly a way to make it more accurate.

 

Yes because a linear model would not work. Ideally, I would want a linear model but when projected across the entirety of the data; both Coalitions will see a net gain for the next infinity. Anything higher than a cubic would produce even steeper prediction results. I suppose I could have used a logarithmic trend, but like the linear model, it would show a net gain in NS for both coalitions when paired with the rest of the data.

Yo man. Exponential/Root functions.

Assume that Y = A + B*C^X.

Y can be the NS, A and B are some arbitrary numbers you find out for me. C is gonna be less than 1, because you've got a decrease every day. X is the number of days the war's been going on.

Congratulations you can now graph under the assumption that each day both sides will consistently X% of what they had the prior day, which is significantly less bananas than cubic functions.

[Do Not Fear Jazz]

• 
  
• Which is why your line graph has only 6 distinct segments instead of 26 right? And unless I missed something, you manually added up the NS totals for both coalitions alliance by alliance? And I thought you said you got your numbers from the wiki?
• That's what I thought. I didn't think you had an answer. I can at least defend my graphs because I use pretty good data and sound procedure. You, on the other hand, pulled your data and prediction right the teat of the Competence propaganda machine.
• Actually you do. From mid-April to the end of June, according to your graph, Competence's NS rises from about 50 million to about 85 million. That is a 14 million per month wartime gain. Holy fuck you guys must be doing something right. And I wasn't referring to your graph when I said the beginning was skewing the data. I was referring to mine and the use of the quadratic equation for a trend line. It is apparent that you don't know what that is so here is an explanation:
• Just because I said I wasn't a math man doesn't mean I don't know how to do it nor does it mean I didn't pass high school geometry. I'm not trying to hide behind "phrases." Just because you don't know what it is, doesn't mean I'm trying to be over complex and misleading. A polynomial trend is also known as "polynomial regression."
  
  "The goal of regression analysis is to model the expected value of a dependent variable y in terms of the value of an independent variable (or vector of independent variables) x."
  
  In instances where the rate of change is not constant (as in our case), a linear polynomial line will not do. A quadratic model can instead be used, and has been used (but in our case, a quartic model was used) to serve as a more accurate trend indicator. The equation I gave you is the equation for the order 4 polynomial line in my graph. That is the equation and there is no other equation possible for a 4 order polynomial for the Sanction Race data.
• That does not explain the graph dropping like that. And, again, how did you determine what alliances drop out? And yes, it is a multi front war, but there is not much interfront fighting. I, for example, have no beef with Umbrella or TOP. If my alliance should surrender to CnG, we would be finished with the war.

 

You shouldn't be using geometry to make graphs.

[headsupsevenup]

I have here some completely unbiased graphs that need to be shared with the rest of the game population. Please review them and understand why goons will win. 

 

 

I'd just like to point out that this is the only graph in this thread that is 100% accurate, verifiable AND reliable.

Edited February 22, 2013 by headsupsevenup

[lamuella]

I like that the axes aren't at zero, implying it is possible to have negative one sixth of a leader deleted.

[crypticedge]

I like that the axes aren't at zero, implying it is possible to have negative one sixth of a leader deleted.

 

That's for new leaders being born, rising and joining the war. I project that function will be used in around 13 years or so though at the soonest, should eQ survive that long. Recreating claiming it was a mistake doesn't count.

[Chad]

You shouldn't be using geometry to make graphs.

Umm...what? lol. Maybe your education system is a little different, but I learned this stuff in freshman year geometry. Perhaps you would have learned it in Algebra or Alg 2, but nonetheless, yes. You use geometry/algebra in graphs. Which is why I have issue with you pulling numbers and trends out of thin air.
 

Yo man. Exponential/Root functions.
Assume that Y = A + B*C^X.
Y can be the NS, A and B are some arbitrary numbers you find out for me. C is gonna be less than 1, because you've got a decrease every day. X is the number of days the war's been going on.
Congratulations you can now graph under the assumption that each day both sides will consistently X% of what they had the prior day, which is significantly less bananas than cubic functions.

I'll admit it is less nuts than a cubic (especially the further out you go), but the problem is that when graphed with the initial spike in NS at the beginning of the war, it doesn't work. For the example I used, however, I'll admit I made a mistake. A cubic poly was not the best fit and the log was a much better prediction model (I used the R-squared value the main reasoning behind which was better among the reasons I stated before).

Equilibrium: y = -5E+10ln(x) + 5E+11
Competence: y = -4E+10ln(x) + 5E+11

And if all of this tells us anything, its that any prediction with NS actually going up is subjective and therefore inaccurate (unless Do Not Fear Jazz can explain what methodology he used to exactly determine why Competence's NS would rise mid-war).

Edited February 23, 2013 by Isotope

[Beets]

I'll admit it is less nuts than a cubic (especially the further out you go), but the problem is that when graphed with the initial spike in NS at the beginning of the war, it doesn't work.

The spike is an outlier caused by a one-time hulk out. Don't start the graph until after the spike, because there's no function in the universe that will account for that cleanly while still being accurate.

[Chad]

The spike is an outlier caused by a one-time hulk out. Don't start the graph until after the spike, because there's no function in the universe that will account for that cleanly while still being accurate.

Which is why I decided to start the graph in February for the example. But I admit you are correct, the poly was not the best prediction model. A log or root function would have been a better fit. I was using the R squared value as an indicator (.96 vs. .99). Mah bad.

Edited February 23, 2013 by Isotope

[lamuella]

The spike is an outlier caused by a one-time hulk out. Don't start the graph until after the spike, because there's no function in the universe that will account for that cleanly while still being accurate.

Exactly this.  The spike in NS is essentially caused by a change in circumstances (nations beefing up for, and starting fighting, the war).  What you need to be graphing is the consequences of continued behaviour, namely what will happen to NS if people keep fighting and if the fight keeps going the way it has.

[Chad]

Exactly this.  The spike in NS is essentially caused by a change in circumstances (nations beefing up for, and starting fighting, the war).  What you need to be graphing is the consequences of continued behaviour, namely what will happen to NS if people keep fighting and if the fight keeps going the way it has.

Wow you're annoying. First off, I've already fix it. Thank you for quoting Beets and thinking so well for yourself. Second, the spike is not caused by nations beefing up for war, the spike was caused by alliances joining in. While I thank you for your unneeded input, Beets' posts don't need a carbon copy commentary.

Edited February 23, 2013 by Isotope

[Sniper Joe]

Guys I made a graph too

 

[Beets]

I took the liberty of summing up all the points that have been brought up in this thread, and came up with this.

If anyone needs clarification on something, gimme a holler
 

[Chad]

Guys I made a graph too

 

If I am reading this correctly; as time increases, more people care? Right on, man. Thank you very much :)

[crypticedge]

If I am reading this correctly; as time increases, more people care? Right on, man. Thank you very much :)

He's saying as more posts get made only math nerds and intentionally awful posters care. I know I'm intentionally awful, as are you (since your so bad at this)

[Chad]

No no no. The graph clearly shows that as time increases, the level of "who cares" rises. Awful posts rise, yes, until a certain point in time it ends. The number of posts by math nerds also rises until a certain amount of time. When math nerds and voices of reason stop posting, the level of dumbness and stupidity sky rockets (no doubt because the only ones left are you and the others attempting to derail the thread).

Edited February 23, 2013 by Isotope

[Puddles]

No no no. The graph clearly shows that as time increases, the level of "who cares" rises. Awful posts rise, yes, until a certain point in time it ends. The number of posts by math nerds also rises until a certain amount of time. When math nerds and voices of reason stop posting, the level of dumbness and stupidity sky rockets (no doubt because the only ones left are you and the others attempting to derail the thread).

The level of who cares rises between the math nerds and awful posters, not everyone else.

 

 

hope that helps

[Chad]

The level of who cares rises between the math nerds and awful posters, not everyone else.

 

 

hope that helps

And apparently, in time, the dumb rest of you will care too :)

[Sniper Joe]

No no no. The graph clearly shows that as time increases, the level of "who cares" rises. Awful posts rise, yes, until a certain point in time it ends. The number of posts by math nerds also rises until a certain amount of time. When math nerds and voices of reason stop posting, the level of dumbness and stupidity sky rockets (no doubt because the only ones left are you and the others attempting to derail the thread).

 

The blue line just entered the stratosphere. Soon it will conquer the galaxy.
 

[crypticedge]

No no no. The graph clearly shows that as time increases, the level of "who cares" rises. Awful posts rise, yes, until a certain point in time it ends. The number of posts by math nerds also rises until a certain amount of time. When math nerds and voices of reason stop posting, the level of dumbness and stupidity sky rockets (no doubt because the only ones left are you and the others attempting to derail the thread).

 

You once again show you don't understand graphs.

[Neo Uruk]

No no no. The graph clearly shows that as time increases, the level of "who cares" rises. Awful posts rise, yes, until a certain point in time it ends. The number of posts by math nerds also rises until a certain amount of time. When math nerds and voices of reason stop posting, the level of dumbness and stupidity sky rockets (no doubt because the only ones left are you and the others attempting to derail the thread).

You are all three lines at once. Congrats.