IA Type II -- Population Trends in China

[dessskris]

plot the data in your GDC, generate the best logistic regression. or use spreadsheet, do some data analysis thing to generate the logistic regression.

[Doreo]

Hi!

I'm in the second question of this IA :/ When they ask for the types of functions that could model the behaviour of the graph?

I got a linear graph :/ (as Population is directly proportional to time (years)) so is that the only function that can model it? OR what?

EDIT:

Why can it be modelled with every function? :/

Edited September 25, 2011 by Doreo

[dessskris]

haha definitely not linear if it was just linear, it wouldn't have been in an IA!

not all functions could model the data accurately. like, not linear and quadratic. not exponential or logarithmic (although they're pretty close).

many people say it's 6th polynomial. but there are some other functions you can model that will fit the data accurately.

[Doreo]

Cubic looked quite right

and why sixth polynomial? :/

Edited September 25, 2011 by Doreo

[dessskris]

nooo it's not accurate enough! try to plot the data in Ms.Word or Excel and try to generate polynomial regressions. which one do you think looks best?

[Doreo]

nooo it's not accurate enough! try to plot the data in Ms.Word or Excel and try to generate polynomial regressions. which one do you think looks best?

Yea 6t polynomial looks right!

But do I actually say I used Excel to find out? What explanation do I give to it?

And then when I analytically develop a model do I simply use x^6 + x^5 + x^4 + x^3 + x^2 + x^1 + x^0?

[dessskris]

lol. I don't think I could give you that much of a guidance!

look at the shape created by the data. does it go up up up down down down (which means you'll need to use quadratic), or up up down down up up (which means you'll need to use cubic), or up down up down up down (which you need to think what you should use). of course you'll need to explain it in a more sophisticated way, but I can't help you with that (my English sucks anyway so I won't be much of help in this part).

and, the general formula for 6th polynomial would be:

y = ax^6 + bx^5 + cx^4 + dx^3 + ex^2 + fx + g

where a,b,c,d,e,f,g are constants.

[Doreo]

Alright thank you so much!

Yea I understood why it's sixth

Thanks once again!

[mbeaumont999]

"Analytically develop one model function that fits the data points on your graph."

At this point, do I solve a simultaneous equation (example: for a sextic function) or would I look at a generic function then transform this so that it can fit my data points?? Not quite sure what to do at this point as equal numbers of the class seem to be doing either one or the other?

[Sabs44]

Hey everyone, just a quick question:

Is going to a ninth degree polynomial (nonic) too far? I can easily calculate it, but will it be better or worse than sticking with a 5th or 6th degree?

Thanks in advance!

Sam

(please quote me with answers)

(and delete this post if I'm not allowed to ask this sort of thing haha)

[dessskris]

At this point, do I solve a simultaneous equation (example: for a sextic function)

Yes

or would I look at a generic function then transform this so that it can fit my data points??

No. This is for the last part, the IMF data.

[dessskris]

Hey everyone, just a quick question:

Is going to a ninth degree polynomial (nonic) too far? I can easily calculate it, but will it be better or worse than sticking with a 5th or 6th degree?

Thanks in advance!

Sam

(please quote me with answers)

(and delete this post if I'm not allowed to ask this sort of thing haha)

it's your choice. just remember you need to justify why you chose that function.

it's fine to use 9th polynomial as long as you can justify its use.

[calito]

have you guys already finished it??

[CkyBlue]

I have done the actual portfolio almost a year ago.

^and to address that, yes, as long as you can justify a nonic function But I recommend you have really good reasons for such a high degree curve... Normally, I would not do it unless it very accurate models it. Remember again, it is not the function goes through all the points, it is the function that models the data considering the nature of what is being analysed, and considering what is reasonable data. Of course, there is no perfect model, so you may want to consider explaining a balance of modelling it and the function going through points with regards to your choice.

[Sabs44]

I understand what you mean... but I have a nonic function with an r^2 of 1, so there's no need to change it...

[CkyBlue]

That doesn't mean it's the perfect model. Would you have a population function that goes through all the points, and then suddenly drop to the negatives right after or in the long run?

[Sabs44]

Oh I completely agree it's unrealistic for future and past, but so are basically all of the functions, other than logistic :/

[Bradley]

Arctan is a sigmoidal shape as well, maybe use that?

[dessskris]

Brad please don't give out advice like that unless you've already completed the portfolio like I have.

I did it in my spare time for fun and got 20/20.

mughtymuddy please don't talk here like that unless you are willing to help people complete the portfolio like I am.

I did it in my busy time because I enjoy helping people and I don't have the guts to leave them struggling; and have helped people get good marks.

[Olga Gonzalez]

I am using a cubic function too. I already substituted each of the coordinates in the cubic model (ax + bx + cx + d....etc etc (imagine it with the exponents)) haha you get the idea. And i was thinking of using a matrix to obtain abcd...the problem is..I don´t really know how to use a matrix to do that. Any help?