Stats help

il0seonpurpose

Korea (South)5638 Posts

September 26 2008 05:23 GMT

I'm doing a project on michael jordan, total points vs his age, and I needed to do two quantitative variables. the problem is though, theres techinically one and a half ages he goes through per season, ill give an example.

Age Year in the NBA PPG
21-22 1984-1985 28.2
22-23 1985-1986 22.7
23-24 1986-1987 37.1
24-25 1987-1988 35
25-26 1988-1989 32.5
26-27 1989-1990 33.6
27-28 1990-1991 31.5
28-29 1991-1992 30.1
29-30 1992-1993 32.6
30-31 Retired 0
31-32 1994-1995 26.9
32-33 1995-1996 30.4
33-34 1996-1997 29.6
34-35 1997-1998 28.7
35-36 Retired 0
36-37 Retired 0
37-38 Retired 0
38-39 2001-2002 22.9
39-40 2002-2003 20

So basically, I have to compare these, and the age is a problem because theres two ages, and I don't know the ppg for each individual year, b/c each year overlaps each other in seasons. this is probably confusing so please if you need, ill clarify, and please help!!

tiffany

3664 Posts

September 26 2008 05:59 GMT

just choose to start counting from either the left number or the right number, and stick with it throughout. you can even maintain a statistical approach in your decision; which age/year made up a higher percentage of the season? you can assume that he scored the majority of his points during the second half (right side year) of the season though, since the all-star game - which indicates the halfway point - is in mid february. unfortunately his birthday is february 17, which is pretty much exactly halfway through the season.

but i think if you just choose one setting and stick with it you'll be fine.

il0seonpurpose

Korea (South)5638 Posts

September 26 2008 06:33 GMT

Yeah, thanks. you know alot about basketball! and also he would score more cuz he went to the playoffs and such

economist_

Vietnam719 Posts

September 26 2008 07:15 GMT

If you are deciding year to be your level of analysis then you may have to determine rigorously which year that his performance belongs to. This could be done by assumptions that he is likely to be belonging to certain period based upon the points. You could also do some sort of extrapolation or intrapolation to do that but it is quite comlicated
It would be perfect you can have the more disggregated level of time series uinit, like quarter of month, season or so. But if you cant, definitly you have to figure out the point that is corresponding to each year. Like in football, stats are always classified by seasons, you would also collect the same thing ideally.
Hope this helps

tiffany

3664 Posts

September 26 2008 07:53 GMT

On September 26 2008 15:33 il0seonpurpose wrote:
Yeah, thanks. you know alot about basketball! and also he would score more cuz he went to the playoffs and such

most places record separate stats for regular season and playoffs. so most likely playoffs won't have factored into those averages. i was thinking more along the lines of: since the halfway point is in mid february, there are technically more games played in the right side year than the left side, and thus more points scored (most likely).

also i have read economist's reply three times and i still have no clue what he is trying to say

economist_

Vietnam719 Posts

September 26 2008 08:33 GMT

Nah, it is age , I miss that, i thought it was only year. Age vs points which are recorded by season? Tough enough. It is like you have age, schooling results and schooling years that not correspodning to age. I would normally suggest that you convert year into age.It is easy in education analysis but really i dont know how to apply to basketball. Tiffany may help ^^.

I am writing this after drinking 2 cans of beer but now i am at office working so hope this helps

Zalfor

United States1035 Posts

September 26 2008 08:46 GMT

i would try a parabolic least squares approximation

or a best fit parabola for that, it seems right.

Please or register to reply.

Stats help

Completed

Ongoing

Upcoming