11 Dec 2010, 1:30 am
11 Dec 2010, 10:09 pm
lotuspuppy
Veteran
Joined: 14 Jan 2008
Age: 36
Gender: Male
Posts: 995
Location: On a journey to the center of the mind
13 Dec 2010, 9:39 pm
04 Jan 2011, 12:45 am
I never took statistics, but I started listening to statistics on KhanAcademy, and since then I've been analysing huge data sets in excel for funsies 
I'm planning on fiddling with the human genome soon, but I think 3 x 10^9 data points might be a bit of a strain on my PC
04 Jan 2011, 1:11 am
ryan93 wrote:
I've been analysing huge data sets in excel for funsies
You should never use Excel for stats; it has a lot of basic errors in how it performs statistical tests and Microsoft refuses to fix them. I've even seen it produce negative r^2 values.
For real statistical work, you should use the R programming language.
If you want a spreadsheet program for playing with data, I would recommend Gnumeric.
They're both free, so you can get them and play with them without paying any money.
04 Jan 2011, 1:53 am
Quote:
For real statistical work, you should use the R programming language.
I've played with R once or twice, but I find it too unintuitive. What I like about excel is that in one column you can can have a data set X, in another you have F(x), and so one. You can compose functions in a very intuitive way, and it's easy to filter and manipulate data according to it's address. It's easy to manipulate one number in R, but data sets are impossible without using spreadsheet, at least for an uneducated freshman like me
Gnumerics does look promising though, thanks a lot for the recommendation. I just hope it's not lacking too many features
04 Jan 2011, 11:42 am
ryan93 wrote:
I've played with R once or twice, but I find it too unintuitive. What I like about excel is that in one column you can can have a data set X, in another you have F(x), and so one. You can compose functions in a very intuitive way, and it's easy to filter and manipulate data according to it's address. It's easy to manipulate one number in R, but data sets are impossible without using spreadsheet, at least for an uneducated freshman like me
R is a very different paradigm; it is a whole programming language after all. The learning curve is steep, but it is worth it.
Quote:
Gnumerics does look promising though, thanks a lot for the recommendation. I just hope it's not lacking too many features
Features are implemented differently than in Excel, so it may take some getting used to, but nothing important is missing. There's even a lot of cool stuff in there that Excel doesn't have.
