Re^2: formula creating algorithm

I never knew linear regression before, I do have some experience with quantitative finance though its often hard to find data sources apart from html parsing yahoo finance archives. I did not know i was going to get such a fast response, if other people can contribute i would be very grateful to hear their advice or maybe some handy perl module recommendations as well but this is already such a helpful answer. I thank you very much.

Comment on Re^2: formula creating algorithm

Replies are listed 'Best First'.
Re^3: formula creating algorithm by BrowserUk (Patriarch) on Jul 06, 2012 at 22:06 UTC
maybe some handy perl module Take a look at Statistics::Regression. Not a "recommendation" as I've never used it, but it might get you started. It is also possible, and often more effective and efficient, to tackle this kind of problem using Genetic Programming algorithms. But you need to know pretty well what conclusions you are looking for up front to construct the algorithm, and that often leads to the curse of statistics: the forgone conclusion. Ie. Finding exactly what you set out to find. But done properly -- using independent criteria -- it can reach conclusion -- positive ot negative -- very quickly. Which can be very useful for exploring a whole gamut of possible correlations and rejecting many of them as early as possible. You can then concentrate the usually slower, more laborious, formal arithmetic techniques on the few remaining. With the rise and rise of 'Social' network sites: 'Computers are making people easier to use everyday' Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error. "Science is about questioning the status quo. Questioning authority". In the absence of evidence, opinion is indistinguishable from prejudice. The start of some sanity?	[reply]

Replies are listed 'Best First'.

Re^3: formula creating algorithm
by BrowserUk (Patriarch) on Jul 06, 2012 at 22:06 UTC

maybe some handy perl module

Take a look at Statistics::Regression. Not a "recommendation" as I've never used it, but it might get you started.

It is also possible, and often more effective and efficient, to tackle this kind of problem using Genetic Programming algorithms.

But you need to know pretty well what conclusions you are looking for up front to construct the algorithm, and that often leads to the curse of statistics: the forgone conclusion. Ie. Finding exactly what you set out to find.

But done properly -- using independent criteria -- it can reach conclusion -- positive ot negative -- very quickly. Which can be very useful for exploring a whole gamut of possible correlations and rejecting many of them as early as possible. You can then concentrate the usually slower, more laborious, formal arithmetic techniques on the few remaining.

With the rise and rise of 'Social' network sites: 'Computers are making people easier to use everyday'

Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error.

"Science is about questioning the status quo. Questioning authority".

In the absence of evidence, opinion is indistinguishable from prejudice.

The start of some sanity?

[reply]