I have the summer off and as such I’m looking for some books that can help guide me doing some actual regression analysis.
One of the most interesting parts of the CFA L2 curriculum for me was multiple regression, and there’s all sorts of stuff I’d like to try it out on (particularly baseball, because I am a big fan and there’s lots of freely available data).
However, the CFA curriculum gives you the theory but I’m certainly not ready right now to run a good regression analysis. Does anyone have suggestions for books or textbooks that walk you through different regression analyses where I can actually do the analysis alongside the book? Something where data sets are available for download and you can physically manipulate the data you’re working with.
tickersu suggested a couple texts for me but I was only able to find them for sale and they were pretty expensive. I’m willing to spend money but I’m hoping to find something in my library if at all possible. It doesn’t have to be baseball themed or anything…but…triple points if anybody can suggest a sabermetrics book that actually goes into dirty detail into the regression required for many of the advanced stats in the game. Most of the baseball books I’ve found are not technical enough. They explain WAR, wrc+, xFIP, etc… but they just sort of tell you that you have to take their word on how the regression equations are developed.
Sorry to hear they weren’t available from a local library… I can tell you I got relatively lucky for how much I paid for each book, but they are good if you can get your hands on them (particularly the one I mentioned to you about applied regression–it has walk trough case studies [data included with the book] as you mentioned, just not necessarily baseball examples). There might be a regression text for sports, but I can only recommend (comfortably) what I’ve used. You can also check out stack exchange for statistics and see if anyone has any suggestions. I hope you’re able to find something that’s cost effective and interesting for you!
I forgot to mention that previous (and or second hand) versions of these books (most texts) are usually much cheaper and generally work just as well (unless it’s accounting and the book is outdated within 5 months…)
What are you meaning by multivariate… Multiple regression or MANOVA, MANCOVA, Factor analysis, PCA (etc.)?
Either way, I agree excel will be pretty weak for many things…SPSS is decent but we can use something with (way) more than enough horsepower in SAS.(Except, SAS is incredibly expensive.)
These use Eviews to run regressions. Eviews spits out many statistics that you cover in L2. I like it because its really easy to run automatic filters to smooth out seasonality. You can also forcast with standard errors. Not SAS, but very easy to use.
Yes, amongst others I was refering to those methods. Depending on the question/hypothesis you want to analyze a simple linear regression model will not be sufficient.
Yes, I’m aware-- I’ve done a bit with those and other multivariate techniques. I think he is just looking to start on regression, but either way, I think excel is a little weak or cumbersome at best. You can certainly start with it for basic ideas and practice.
Before going into theory, you will have to decide on software first. Excel is very limited to simple OLS (there are some add-ons, but they are either very costly or hardly reliable). EViews and Stata are very newbie-friendly solutions, but R is still the way to go for any serious statistical analysis. There is a vast community around R and tons of books on various special topics.
If you are comfortable with matrix algebra, Greene’s Econometric Analysis, mostly considered the bible of econometrics (with more than 1200 pages a fitting analogy), is the onyl way to go and will get you from anything OLS to GLM and fixed/random effects models.
Woolridge is a good intro, if your matrix algebra is rusty. To practice, you will need some software guided book. I recommend Kleiber & Zeileis’ Applied Econometrics in R as a practical intro to regression analysis.
Once you are comfortable with the above, the next logical step is Tsay’s Analysis of Financial Time Series, which will get you where I guess you will really want to go - the analysis of finance data. Once you’re done with that I have a whole plethora of follow-up readings for you if you are still interested then.
From what I’ve heard, Mostly Harmless Econometrics is a good book, but is very narrow in scope-- it leaves out time series and several topics in a common regression course- interaction terms, for example (but I haven’t used the book, so I can’t verify this). At some point, you can just jump in and start. Eventually, you will learn more if you keep digging…