(ns code.pls.ncliff
  (:use (code.pls cliff))
  (:use (code.utils utils preprocess)) 
  (:use (incanter core stats)))

(defn ncliff1 [n data _ deci-n _]
"Returns a pruned list of instances. Designed
 for discrete data.

 data=original data, n=number of bins,
 deci-n=% instances selected
"  
  (let [cliff-data (extract-unique
		    (select-numeric-instances
		     data
		     deci-n))]
    cliff-data))

(defn ncliff [n data1 _ deci-n _]
"Returns a pruned list of instances. Designed
 for numeric data.

 data=original data, n=number of bins,
 deci-n=% instances selected
"  
  (let [data (bind-columns (sel data1 :cols (range 10)) (matrix (efb (sel data1 :cols 10))) (sel data1 :cols (range 11 21)))
  			klass (sel data :cols (dec (ncol data)))
	ind-vars (sel data :cols (range (dec (ncol data))))
	discretized-data (bind-columns
			  (trans (efb2 ind-vars n))
			  klass)
	cliff-data (extract-unique
		    (select-numeric-instances
		     discretized-data
		     deci-n))
	match-data (fn [dat]
		     (loop [n 0
			    dd discretized-data
			    d data
			    result (transient [])]
		       (if (empty? d)
			 (persistent! result)
			 (recur
			  (inc n)
			  (rest dd)
			  (rest d)
			  (conj! result (if (= dat (first dd))
					 (first d)
					 nil))))))
	result (apply concat (map #(match-data %) cliff-data))]  
    (extract-unique (matrix (remove #(= nil %) result)))))