(ns code.preprocessors.utils.bestk
	(:use (code.learners.knn knn_classifier)) 
  (:use (code.utils utils folds))
  (:use (code.experiments knn prep))
  (:use (incanter core stats)))
(declare superk bestk1)

(defn bestk [n x y distance]
"n = 5, x = 10, y = data number"
	(let [one (apply concat (map #(bestk1 n x % distance) (second (nth cdatasets y))))
				two (apply min (map first one))
				bk (second (first (sort-by second (filter #(= (first %) two) one))))]
		bk))				
		

(defn bestk1 [n x D distance]
  (let [pro-folds (folds D n)]
    (loop [fold pro-folds result []]
      (if (empty? fold)
        result
        (recur (rest fold)
               (conj 
               	result
		(let [training (:train (first fold))]
			(superk training x distance))))))))		      

(defn superk [D x distance]
	(let [dat D
				N (take x dat)
				TN (loop [d D result []]
							(if (empty? d)
								(remove #(= 'none %) result)
								(recur (rest d)
											 (conj result (if (member? (first d) N)
											 								'none
											 								(first d))))))
				bk (loop [k (range 1 (+ 1 (count TN))) result []]
							(if (empty? k)
								(first (sort-by first result))
								(recur (rest k)
											 (conj result (let [got (map #(knn-classifier (butlast %) (first k) TN distance) N)
											 										want (last (Transpose N))
											 										goals (unique-goals D)
											 										pf (median (map second (abcd-stats want got goals)))]
											 									[pf (first k)])))))]
			bk))