gistfile1.clj

(ns ml-jam.core
  (:use clojure-csv.core))

(def training-data
  (for [row (rest (parse-csv (slurp "./digitssample.csv")))]
    (for [x row] ( Integer/parseInt x))))

(def check-data
  (for [row (rest (parse-csv (slurp "./digitssample.csv")))]
    (for [x row] ( Integer/parseInt x))))

(defn distance [r1 r2]
   (reduce + (map (fn [x y] (Math/pow (- x y) 2)) (rest r1) (rest r2)))) 

(defn class-of [r]
  (first r))

(defn classify [row tset]
  (let [groups (group-by first tset)
        averages (pmap (fn [[k v]] [k (/ (reduce + (mapv (partial distance row) v)) (count v))]) groups)
        ;;averages (for [[k v] groups]
        ;;              [k (/ (reduce + (mapv (partial distance row) v)) (count v))])
       ]
      (first (sort-by second averages))))

(defn check [data]
  (for [r data]
    [(first r) (classify r training-data)]))

(defn results [input]
  (let [result (check input)]
    (println (float (/ (count (filter #(= (first %) (first (second %))) result)) (count result))))))

;;  (map (fn [x y] (Math/pow (- x y) 2)) (nth training-data 1) (nth training-data 2))
;;  (map (fn [x y] (println x)) (nth training-data 0) (nth training-data 1))
;;  (distance (nth training-data 0) (nth training-data 1))
;;  (classify (first training-data) training-data)
;;  (time (results (take 10 check-data)))