@@ -25,7 +25,7 @@ ([vectors nodes] (let [clusters (loop [index 0 - clusters (list-of (count nodes) [])] + clusters (vec (replicate (count nodes) []))] (if (= (count vectors) index) clusters (let [vector (nth vectors index) @@ -48,7 +48,7 @@ The return value will be a tree of Maps of the form: { :vec [] :left { :vec ... } :right { :vec ... } }" [nodes] - (if (= 1 (count nodes)) + (if (< (count nodes) 2) nodes (let [vectors (map #(get % :vec) nodes) [closest-pair cls] (closest-vectors vectors) cluster.clj @@ -9,14 +9,22 @@ (defn sum [v] (reduce + v)) -(defn pearson [v1 v2] - (let [v1mean (mean v1) - v2mean (mean v2)] - (/ (sum (map - #(* (- %1 v1mean) (- %2 v2mean)) - v1 v2)) - (sqrt (* (sum (map #(square (- % v1mean)) v1)) - (sum (map #(square (- % v2mean)) v2))))))) + + +(defn pearson [x y] + (let [n (count x) + o (/ 1 n) + sum-y (sum y) + sum-x (sum x) + prod-of-sqrts (* (sqrt (- (sum (map #(square %) x)) + (* o (square sum-x)))) + (sqrt (- (sum (map #(square %) y)) + (* o (square sum-y)))))] + (if (= 0.0 prod-of-sqrts) + 0.0 + (/ (- (sum (map #(* %1 %2) x y)) + (* o sum-x sum-y)) + prod-of-sqrts)))) (defn compact [v] (loop [v1 v out '()] @@ -98,11 +106,6 @@ (recur (dec how-many-left) (cons (random-vector length range-start range-end) vectors))))) -(defn list-of [how-many value] - (loop [how-many-left how-many - list []] - (if (= 0 how-many-left) - list - (recur (dec how-many-left) (conj list value))))) + internal/internal.clj b85bfb1a0959f5607f6afc73d77fc60172b92652 Tyler McMullen tbmcmullen@gmail.com http://github.com/tyler/clojure-cluster/commit/92b557ae578889674c8f806085dcb9e3daf04c1c 92b557ae578889674c8f806085dcb9e3daf04c1c 2008-11-06T23:40:33-08:00 2008-11-06T23:40:33-08:00 Faster version of Pearson and use replicate not list-of. b0b867789ee0603fa3ffd4e08cb19ebdab6ad277 Tyler McMullen tbmcmullen@gmail.com