Comments on: Ranks Of, Ordered Ranks Of and Tied Ranks Of (Statistics in Erlang part 4) http://fullof.bs/ranks-of-ordered-ranks-of-and-tied-ranks-of-statistics-in-erlang-part-4/ He just never stops talking Fri, 16 Dec 2011 00:30:46 +0000 hourly 1 http://wordpress.org/?v=3.2.1 By: Recent Links Tagged With "correlations" - JabberTags http://fullof.bs/ranks-of-ordered-ranks-of-and-tied-ranks-of-statistics-in-erlang-part-4/comment-page-1/#comment-876 Recent Links Tagged With "correlations" - JabberTags Sun, 30 Nov 2008 12:49:20 +0000 http://fullof.bs/?p=323#comment-876 [...] How Correlation Between Asset Classes Affects Your Portfolio Saved by OGDIABLO on Tue 11-11-2008 Ranks Of, Ordered Ranks Of and Tied Ranks Of (Statistics in Erlang ... Saved by socialmediamarketing on Mon 10-11-2008 Physicists Seek Answers to Quantum Correlations [...] [...] How Correlation Between Asset Classes Affects Your Portfolio Saved by OGDIABLO on Tue 11-11-2008 Ranks Of, Ordered Ranks Of and Tied Ranks Of (Statistics in Erlang … Saved by socialmediamarketing on Mon 10-11-2008 Physicists Seek Answers to Quantum Correlations [...]

]]> By: Spearman’s Rank Correlation Coefficient (Statistics in Erlang part 7) | Full of BS http://fullof.bs/ranks-of-ordered-ranks-of-and-tied-ranks-of-statistics-in-erlang-part-4/comment-page-1/#comment-875 Spearman’s Rank Correlation Coefficient (Statistics in Erlang part 7) | Full of BS Sun, 24 Aug 2008 16:36:33 +0000 http://fullof.bs/?p=323#comment-875 [...] is free and MIT licensed, because the GPL is evil.  This code uses the ordered ranks code from Part 4.  This closes issue [...] [...] is free and MIT licensed, because the GPL is evil.  This code uses the ordered ranks code from Part 4.  This closes issue [...]

]]> By: Statistical Correlations (Statistics in Erlang part 5) | Full of BS http://fullof.bs/ranks-of-ordered-ranks-of-and-tied-ranks-of-statistics-in-erlang-part-4/comment-page-1/#comment-874 Statistical Correlations (Statistics in Erlang part 5) | Full of BS Sat, 23 Aug 2008 22:45:09 +0000 http://fullof.bs/?p=323#comment-874 [...] way this is achieved is through ranking, which was covered in Statistics in Erlang part 4.  The general idea is straightforward: just make a list of your values’ ranks from most [...] [...] way this is achieved is through ranking, which was covered in Statistics in Erlang part 4.  The general idea is straightforward: just make a list of your values’ ranks from most [...]

]]> By: John Haugeland http://fullof.bs/ranks-of-ordered-ranks-of-and-tied-ranks-of-statistics-in-erlang-part-4/comment-page-1/#comment-872 John Haugeland Fri, 15 Aug 2008 15:27:56 +0000 http://fullof.bs/?p=323#comment-872 Yeah, sorts in typable languages are generally o(lg n) with a large constant, whereas lists:reverse is just a set of pointer swaps, which most CPUs have atomic ops for; avoiding sort where possible is awesome. Still, I can probably take some traversals out this weekend, when I get down to actually thinking about it. :D Yeah, sorts in typable languages are generally o(lg n) with a large constant, whereas lists:reverse is just a set of pointer swaps, which most CPUs have atomic ops for; avoiding sort where possible is awesome.

Still, I can probably take some traversals out this weekend, when I get down to actually thinking about it. :D

]]> By: Anders Nygren http://fullof.bs/ranks-of-ordered-ranks-of-and-tied-ranks-of-statistics-in-erlang-part-4/comment-page-1/#comment-873 Anders Nygren Fri, 15 Aug 2008 13:11:09 +0000 http://fullof.bs/?p=323#comment-873 After actually timing my previous code I found that lists:sort/2 is very slow. In this case it is actually better to do lists:reverse(lists:sort(List)) After actually timing my previous code I found that lists:sort/2 is very slow.
In this case it is actually better to do lists:reverse(lists:sort(List))

]]> By: Anders Nygren http://fullof.bs/ranks-of-ordered-ranks-of-and-tied-ranks-of-statistics-in-erlang-part-4/comment-page-1/#comment-871 Anders Nygren Fri, 15 Aug 2008 10:27:43 +0000 http://fullof.bs/?p=323#comment-871 This only traverses the list 2 times in stead of four. <code> ranks_of(List) when is_list(List) -> {L,_} = lists:foldl(fun (E,{L,S}) -> {[{S,E}|L],S+1} end, {[],1}, lists:sort(fun erlang:'>' /2, List)), lists:reverse(L). </code> This only traverses the list 2 times in stead of four.

ranks_of(List) when is_list(List) ->
{L,_} = lists:foldl(fun (E,{L,S}) ->
{[{S,E}|L],S+1}
end, {[],1}, lists:sort(fun erlang:'>' /2, List)),
lists:reverse(L).

]]> By: John Haugeland http://fullof.bs/ranks-of-ordered-ranks-of-and-tied-ranks-of-statistics-in-erlang-part-4/comment-page-1/#comment-870 John Haugeland Fri, 15 Aug 2008 04:29:20 +0000 http://fullof.bs/?p=323#comment-870 Good point. I'll fix that this weekend when I add in Anders Nygren's replacement code and Alain O'Dea's suggestions. Good point. I’ll fix that this weekend when I add in Anders Nygren’s replacement code and Alain O’Dea’s suggestions.

]]> By: vat http://fullof.bs/ranks-of-ordered-ranks-of-and-tied-ranks-of-statistics-in-erlang-part-4/comment-page-1/#comment-869 vat Fri, 15 Aug 2008 04:22:00 +0000 http://fullof.bs/?p=323#comment-869 in tied_add_prev, you're calling length multiple times unnecessarily. if FoundAt is really long, then this could be a very expensive operation. in tied_add_prev, you’re calling length multiple times unnecessarily. if FoundAt is really long, then this could be a very expensive operation.

]]>