Ah yes, fun ol' statistics
So I'm at work and have a couple hours off, and what do I do with my free time? Read about time series and regression!
I found an interesting pdf document (I think it's a book released as part of the GNU project).
Here is the link
As you can see (if you clicked the link), it is published by a bunch of nazis...otherwise known as Germans. All the exercises are crunching data from Germany such as average income in DMs (Deutsche marks) and population growth in different regions of Germany. My undergraduate advisor suggested picking up either German or Chinese for those who wish to pursue studies in higher level mathematics as the nations which speak those languages (Germany and China...if you can imagine that) are mathematics hotbeds. So yes, Germany and China are down wit da numba's yo!
I'm still on chapter 1 as I am working through the different SAS programming exercises. I am learning new functions of SAS as I go. Apparently (as you may or may not read) time series and regression are somehow tied together. Basically, if your independent variable is time (how about discrete increments...say...seconds, months, days, etc.) then you use regression to model the behavior of a system over time. Simple transformations can foster linear behavior and normally distributed error terms pop into the picture. The real fun begins when the error terms aren't the boring ol' normal process, but, instead, distributed according to a stochastic process such as Brownian Motion or (as we learned this semester) a martingale.
I even happened to read earlier today about the regression relationship between the price of a barrel of oil and the price of a gallon of gasoline. They analyzed profit margins (which requires some kind of understanding of finance I guess), but their model was the cool part. It featured a white noise error term. White noise happens to be a stochastic process and I think it is somehow related to time series/regression....well, obviously regression...it was featured as the error terms in the econometricians' models :)
Anyhow, yes, if you are studying engineering, chemistry, or any other discipline which has as part of its core some calculus (probably gonna need about 12-16 hours of it), I encourage you to jump into the wonderful world of statistics. It can be quite an adventure.
I found an interesting pdf document (I think it's a book released as part of the GNU project).
Here is the link
As you can see (if you clicked the link), it is published by a bunch of nazis...otherwise known as Germans. All the exercises are crunching data from Germany such as average income in DMs (Deutsche marks) and population growth in different regions of Germany. My undergraduate advisor suggested picking up either German or Chinese for those who wish to pursue studies in higher level mathematics as the nations which speak those languages (Germany and China...if you can imagine that) are mathematics hotbeds. So yes, Germany and China are down wit da numba's yo!
I'm still on chapter 1 as I am working through the different SAS programming exercises. I am learning new functions of SAS as I go. Apparently (as you may or may not read) time series and regression are somehow tied together. Basically, if your independent variable is time (how about discrete increments...say...seconds, months, days, etc.) then you use regression to model the behavior of a system over time. Simple transformations can foster linear behavior and normally distributed error terms pop into the picture. The real fun begins when the error terms aren't the boring ol' normal process, but, instead, distributed according to a stochastic process such as Brownian Motion or (as we learned this semester) a martingale.
I even happened to read earlier today about the regression relationship between the price of a barrel of oil and the price of a gallon of gasoline. They analyzed profit margins (which requires some kind of understanding of finance I guess), but their model was the cool part. It featured a white noise error term. White noise happens to be a stochastic process and I think it is somehow related to time series/regression....well, obviously regression...it was featured as the error terms in the econometricians' models :)
Anyhow, yes, if you are studying engineering, chemistry, or any other discipline which has as part of its core some calculus (probably gonna need about 12-16 hours of it), I encourage you to jump into the wonderful world of statistics. It can be quite an adventure.