(ns code.privacy.summarize_data
  (:use (code.utils utils))
  (:use (code.learners k_means))
  (:use (incanter core stats)))

(defn summarize [k data distance]
  (let [clusters (k-means-model k data distance)
        summary (map #(mysummary % distance) clusters)]
    summary))

(defn iscovered? [one summary distance idx]
  (loop [s summary result false]
    (if (or (= result true) (empty? s))
      result
      (recur 
        (rest s)
        (if (<= (distance (butlast one) (first (first s))) (second (first s)))
          true
          false)))))

(defn conflict [test-data summary distance idx]
  (let [fight (count (filter #(= idx %) (map #(iscovered? % summary distance idx) test-data)))
        all   (count test-data)]
    (* 100.0 (/ fight all))))

(defn coverage [test-data summary distance idx]
  (* (/ (apply  + (map #(iscovered? % summary distance idx) test-data))
        (count test-data))
     100.0))

(defn new-test [test-data summary distance idx]
  (loop [t test-data result [] train1 []]
    (if (empty? t)
      (let [test (remove #(= 'na %) result)
            train (remove #(= 'na %) train1)]
        [(save (matrix train) "train.csv") (save (matrix test) "test.csv")])
      (recur 
        (rest t)
        (conj result (if (iscovered? (first t) summary distance idx)
                       (first t)
                       'na))
        (conj train1 (if (iscovered? (first t) summary distance idx)
                      'na
                      (first t)))))))