1 ! Copyright (C) 2008 Doug Coleman, Michael Judge.
2 ! See http://factorcode.org/license.txt for BSD license.
3 USING: assocs combinators generalizations kernel locals math
4 math.functions math.order math.vectors sequences
5 sequences.private sorting fry ;
9 [ sum ] [ length ] bi / ;
11 : geometric-mean ( seq -- x )
12 [ length ] [ product ] bi nth-root ;
14 : harmonic-mean ( seq -- x )
15 [ recip ] map-sum recip ;
19 :: (kth-object) ( seq k quot: ( x y -- ? ) nth-quot exchange-quot -- elt )
20 #! Wirth's method, Algorithm's + Data structues = Programs p. 84
21 #! The algorithm modifiers seq, so we clone it
22 k seq bounds-check 2drop
36 [ i seq nth-quot call x quot call ] [ i 1 + i! ] while
37 [ x j seq nth-quot call quot call ] [ j 1 - j! ] while
39 i j seq exchange-quot call
50 : kth-object-unsafe ( seq k quot: ( x y -- ? ) -- elt )
51 [ nth-unsafe ] [ exchange-unsafe ] (kth-object) ; inline
55 : kth-object ( seq k quot: ( x y -- ? ) -- elt )
56 [ nth ] [ exchange ] (kth-object) ; inline
58 : kth-smallest ( seq k -- elt ) [ < ] kth-object-unsafe ;
60 : kth-largest ( seq k -- elt ) [ > ] kth-object-unsafe ;
62 : count-relative ( seq k -- lt eq gt )
65 { +lt+ [ [ 1 + ] 2dip ] }
67 { +eq+ [ [ 1 + ] dip ] }
71 : minmax-relative ( seq k -- lt eq gt lt-max gt-min )
72 [ 0 0 0 -1/0. 1/0. ] 2dip '[
74 { +lt+ [ [ 1 + ] 5 ndip '[ _ max ] dip ] }
75 { +gt+ [ [ 1 + ] 3dip min ] }
76 { +eq+ [ [ 1 + ] 4dip drop ] }
80 : lower-median ( seq -- elt )
81 [ ] [ ] [ length odd? ] tri
82 [ midpoint@ ] [ midpoint@ 1 - ] if kth-smallest ;
84 : upper-median ( seq -- elt )
85 dup midpoint@ kth-smallest ;
87 : medians ( seq -- lower upper )
88 [ lower-median ] [ upper-median ] bi ;
91 [ ] [ length odd? ] bi [ lower-median ] [ medians + 2 / ] if ;
95 : (sequence>assoc) ( seq map-quot: ( x -- ..y ) insert-quot: ( ..y assoc -- ) assoc -- assoc )
96 [ swap curry compose each ] keep ; inline
100 : sequence>assoc! ( assoc seq map-quot: ( x -- ..y ) insert-quot: ( ..y assoc -- ) -- assoc )
101 4 nrot (sequence>assoc) ; inline
103 : sequence>assoc ( seq map-quot: ( x -- ..y ) insert-quot: ( ..y assoc -- ) exemplar -- assoc )
104 clone (sequence>assoc) ; inline
106 : sequence>hashtable ( seq map-quot: ( x -- ..y ) insert-quot: ( ..y assoc -- ) -- hashtable )
107 H{ } sequence>assoc ; inline
109 : histogram! ( hashtable seq -- hashtable )
110 [ ] [ inc-at ] sequence>assoc! ;
112 : histogram-by ( seq quot: ( x -- bin ) -- hashtable )
113 [ inc-at ] sequence>hashtable ; inline
115 : histogram ( seq -- hashtable )
118 : sorted-histogram ( seq -- alist )
119 histogram sort-values ;
121 : collect-pairs ( seq quot: ( x -- v k ) -- hashtable )
122 [ push-at ] sequence>hashtable ; inline
124 : collect-by ( seq quot: ( x -- x' ) -- hashtable )
125 [ dup ] prepose collect-pairs ; inline
129 [ ] [ [ [ second ] bi@ > ] most ] map-reduce first ;
131 ERROR: empty-sequence ;
133 : minmax ( seq -- min max )
137 [ first dup ] keep [ [ min ] [ max ] bi-curry bi* ] each
143 : sample-var ( seq -- x )
148 [ [ mean ] keep [ - sq ] with map-sum ]
156 [ [ mean ] keep [ - sq ] with map-sum ]
160 : std ( seq -- x ) var sqrt ;
162 : ste ( seq -- x ) [ std ] [ length ] bi sqrt / ;
164 : ((r)) ( mean(x) mean(y) {x} {y} -- (r) )
165 ! finds sigma((xi-mean(x))(yi-mean(y))
166 0 [ [ [ pick ] dip swap - ] bi@ * + ] 2reduce 2nip ;
168 : (r) ( mean(x) mean(y) {x} {y} sx sy -- r )
169 * recip [ [ ((r)) ] keep length 1 - / ] dip * ;
171 : [r] ( {{x,y}...} -- mean(x) mean(y) {x} {y} sx sy )
172 first2 [ [ [ mean ] bi@ ] 2keep ] 2keep [ std ] bi@ ;
174 : r ( {{x,y}...} -- r )
177 : r^2 ( {{x,y}...} -- r )
180 : least-squares ( {{x,y}...} -- alpha beta )
181 [r] { [ 2dup ] [ ] [ ] [ ] [ ] } spread
182 ! stack is mean(x) mean(y) mean(x) mean(y) {x} {y} sx sy
183 [ (r) ] 2keep ! stack is mean(x) mean(y) r sx sy
184 swap / * ! stack is mean(x) mean(y) beta
187 : cov ( {x} {y} -- cov )
188 [ dup mean v-n ] bi@ v* mean ;
190 : corr ( {x} {y} -- corr )
191 [ cov ] [ [ var ] bi@ * sqrt ] 2bi / ;
193 : cum-sum ( seq -- seq' )
194 0 swap [ + dup ] map nip ;
196 : cum-product ( seq -- seq' )
197 1 swap [ * dup ] map nip ;
199 : cum-min ( seq -- seq' )
200 [ ?first ] keep [ min dup ] map nip ;
202 : cum-max ( seq -- seq' )
203 [ ?first ] keep [ max dup ] map nip ;