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 ;
17 :: kth-object ( seq k quot: ( x y -- ? ) -- elt )
18 #! Wirth's method, Algorithm's + Data structues = Programs p. 84
19 #! The algorithm modifiers seq, so we clone it
33 [ i seq nth-unsafe x quot call ] [ i 1 + i! ] while
34 [ x j seq nth-unsafe quot call ] [ j 1 - j! ] while
36 i j seq exchange-unsafe
47 : kth-smallest ( seq k -- elt ) [ < ] kth-object ;
49 : kth-largest ( seq k -- elt ) [ > ] kth-object ;
51 : count-relative ( seq k -- lt eq gt )
54 { +lt+ [ [ 1 + ] 2dip ] }
56 { +eq+ [ [ 1 + ] dip ] }
60 : minmax-relative ( seq k -- lt eq gt lt-max gt-min )
61 [ 0 0 0 -1/0. 1/0. ] 2dip '[
63 { +lt+ [ [ 1 + ] 5 ndip '[ _ max ] dip ] }
64 { +gt+ [ [ 1 + ] 3dip min ] }
65 { +eq+ [ [ 1 + ] 4dip drop ] }
69 : lower-median ( seq -- elt )
70 [ ] [ ] [ length odd? ] tri
71 [ midpoint@ ] [ midpoint@ 1 - ] if kth-smallest ;
73 : upper-median ( seq -- elt )
74 dup midpoint@ kth-smallest ;
76 : medians ( seq -- lower upper )
77 [ lower-median ] [ upper-median ] bi ;
80 [ ] [ length odd? ] bi [ lower-median ] [ medians + 2 / ] if ;
84 : (sequence>assoc) ( seq map-quot: ( x -- ..y ) insert-quot: ( ..y assoc -- ) assoc -- assoc )
85 [ swap curry compose each ] keep ; inline
89 : sequence>assoc! ( assoc seq map-quot: ( x -- ..y ) insert-quot: ( ..y assoc -- ) -- assoc )
90 4 nrot (sequence>assoc) ; inline
92 : sequence>assoc ( seq map-quot: ( x -- ..y ) insert-quot: ( ..y assoc -- ) exemplar -- assoc )
93 clone (sequence>assoc) ; inline
95 : sequence>hashtable ( seq map-quot: ( x -- ..y ) insert-quot: ( ..y assoc -- ) -- hashtable )
96 H{ } sequence>assoc ; inline
98 : histogram! ( hashtable seq -- hashtable )
99 [ ] [ inc-at ] sequence>assoc! ;
101 : histogram-by ( seq quot: ( x -- bin ) -- hashtable )
102 [ inc-at ] sequence>hashtable ; inline
104 : histogram ( seq -- hashtable )
107 : sorted-histogram ( seq -- alist )
108 histogram sort-values ;
110 : collect-pairs ( seq quot: ( x -- v k ) -- hashtable )
111 [ push-at ] sequence>hashtable ; inline
113 : collect-by ( seq quot: ( x -- x' ) -- hashtable )
114 [ dup ] prepose collect-pairs ; inline
118 [ ] [ [ [ second ] bi@ > ] most ] map-reduce first ;
120 ERROR: empty-sequence ;
122 : minmax ( seq -- min max )
126 [ first dup ] keep [ [ min ] [ max ] bi-curry bi* ] each
132 : sample-var ( seq -- x )
137 [ [ mean ] keep [ - sq ] with map-sum ]
145 [ [ mean ] keep [ - sq ] with map-sum ]
149 : std ( seq -- x ) var sqrt ;
151 : ste ( seq -- x ) [ std ] [ length ] bi sqrt / ;
153 : ((r)) ( mean(x) mean(y) {x} {y} -- (r) )
154 ! finds sigma((xi-mean(x))(yi-mean(y))
155 0 [ [ [ pick ] dip swap - ] bi@ * + ] 2reduce 2nip ;
157 : (r) ( mean(x) mean(y) {x} {y} sx sy -- r )
158 * recip [ [ ((r)) ] keep length 1 - / ] dip * ;
160 : [r] ( {{x,y}...} -- mean(x) mean(y) {x} {y} sx sy )
161 first2 [ [ [ mean ] bi@ ] 2keep ] 2keep [ std ] bi@ ;
163 : r ( {{x,y}...} -- r )
166 : r^2 ( {{x,y}...} -- r )
169 : least-squares ( {{x,y}...} -- alpha beta )
170 [r] { [ 2dup ] [ ] [ ] [ ] [ ] } spread
171 ! stack is mean(x) mean(y) mean(x) mean(y) {x} {y} sx sy
172 [ (r) ] 2keep ! stack is mean(x) mean(y) r sx sy
173 swap / * ! stack is mean(x) mean(y) beta
176 : cov ( {x} {y} -- cov )
177 [ dup mean v-n ] bi@ v* mean ;
179 : corr ( {x} {y} -- corr )
180 [ cov ] [ [ var ] bi@ * sqrt ] 2bi / ;
182 : cum-sum ( seq -- seq' )
183 0 swap [ + dup ] map nip ;
185 : cum-product ( seq -- seq' )
186 1 swap [ * dup ] map nip ;
188 : cum-min ( seq -- seq' )
189 [ ?first ] keep [ min dup ] map nip ;
191 : cum-max ( seq -- seq' )
192 [ ?first ] keep [ max dup ] map nip ;