(ns code.report.cliff_noise_report
  (:use (code.utils utils preprocess))
  (:use (code.pls cliff ncliff))
  (:use (code.data data))
  (:use (code.defect utils search1 cliff_java))
  (:use (code.privacy privacy))
  (:use (code.report privacy_report))
  (:use (incanter core stats io)))
  
                 
;column 3 in csv is acc
;column 4 in csv is pd
;column 5 in csv is precision                 
;column 6 in csv is fmeasure  
;column 7 in csv is pf
;column 1 in csv is machine learners [5] = [knn] (ml-knn)
;column 2 in csv is training data  (td-knn)
;fm is [1 2 6]             
;group-by column 1 (gb1)
;(extract-acc-data "./code/vote-output-result.csv" [0 2 3] 'acc)
(defn extract-acc-data [result-csv-name a performance-name] ;a for [0 2 3]
	(let [result (report-file result-csv-name) 
				result1 (sel result :cols a)
				gb0 (group-by first result1)
				ml-breastcancer (matrix (second (first (filter #(= 0 (first %)) gb0))))
				ml-ecoli (matrix (second (first (filter #(= 1 (first %)) gb0))))
				ml-glass (matrix (second (first (filter #(= 2 (first %)) gb0))))
				ml-heart-c (matrix (second (first (filter #(= 3 (first %)) gb0))))
				ml-hep (matrix (second (first (filter #(= 4 (first %)) gb0))))
				ml-iris (matrix (second (first (filter #(= 5 (first %)) gb0))))
				ml-labor (matrix (second (first (filter #(= 6 (first %)) gb0))))
				ml-lymph (matrix (second (first (filter #(= 7 (first %)) gb0))))
				ml-vote (matrix (second (first (filter #(= 8 (first %)) gb0))))
				td-breastcancer (group-by #(nth % 1) ml-breastcancer)
				td-ecoli (group-by #(nth % 1) ml-ecoli)
				td-glass (group-by #(nth % 1) ml-glass)
				td-heart-c (group-by #(nth % 1) ml-heart-c)
				td-hep (group-by #(nth % 1) ml-hep)
				td-iris (group-by #(nth % 1) ml-iris)
				td-labor (group-by #(nth % 1) ml-labor)
				td-lymph (group-by #(nth % 1) ml-lymph)
				td-vote (group-by #(nth % 1) ml-vote)
				med-tbl (fn [group-td-data]
							(loop [td (sort-by first group-td-data) result []]
								(if (empty? td)
									result
									(recur
										(rest td)
										(conj result (get-median (last (trans (second (first td))))))))))
										
		;(save (med-tbl td-breastcancer) (str "me-"test-name".dat") :delim \space)))
      	myans	(matrix [(med-tbl td-breastcancer) (med-tbl td-ecoli) (med-tbl td-glass) (med-tbl td-heart-c) (med-tbl td-hep) (med-tbl td-iris) (med-tbl td-labor) (med-tbl td-lymph) (med-tbl td-vote)
		                   ])
		    knn-breastcancer (sel (first myans) :cols [0 1 2 3 4])
		    cliff-breastcancer (sel (first myans) :cols [5 6 7 8 9])
		    knn-ecoli (sel (second myans) :cols [0 1 2 3 4])
		    cliff-ecoli (sel (second myans) :cols [5 6 7 8 9])
		    knn-glass (sel (nth myans 2) :cols [0 1 2 3 4])
		    cliff-glass (sel (nth myans 2) :cols [5 6 7 8 9])
		    knn-heart-c (sel (nth myans 3) :cols [0 1 2 3 4])
		    cliff-heart-c (sel (nth myans 3) :cols [5 6 7 8 9])
		    knn-hep (sel (nth myans 4) :cols [0 1 2 3 4])
		    cliff-hep (sel (nth myans 4) :cols [5 6 7 8 9])
		    knn-iris (sel (nth myans 5) :cols [0 1 2 3 4])
		    cliff-iris (sel (nth myans 5) :cols [5 6 7 8 9])
		    knn-labor (sel (nth myans 6) :cols [0 1 2 3 4])
		    cliff-labor (sel (nth myans 6) :cols [5 6 7 8 9])
		    knn-lymph (sel (nth myans 7) :cols [0 1 2 3 4])
		    cliff-lymph (sel (nth myans 7) :cols [5 6 7 8 9])
		    knn-vote (sel (nth myans 8) :cols [0 1 2 3 4])
		    cliff-vote (sel (nth myans 8) :cols [5 6 7 8 9])]
		[(save (bind-columns (matrix [1 2 3 4 5]) (trans [knn-breastcancer knn-ecoli knn-glass knn-heart-c knn-hep knn-iris knn-labor knn-lymph knn-vote])) (str "./code/report/report-2011-11-03-1/"performance-name"-knn.dat") :delim \space)
		(save (bind-columns (matrix [1 2 3 4 5]) (trans [cliff-breastcancer cliff-ecoli cliff-glass cliff-heart-c cliff-hep cliff-iris cliff-labor cliff-lymph cliff-vote])) (str "./code/report/report-2011-11-03-1/"performance-name"-cliff.dat") :delim \space)]))

;(extract-acc-data2 "./code/report/report-2011-11-18-1/rs3-output-result.csv" [0 2 7] 'pf)
;(extract-performance-data1 "./code/within/ant13-output-result.csv" [1 2 7] 'fm)

(def mylr [1 2 3 4 5])                 
(def mynb [1 2 3 4 5])
(def myrf [1 2 3 4 5])

;(save (bind-columns (range 1 11) (rest (trans mylr))) "./within/lr-all.dat")
;(extract-performance-data1 "./within/ant13-output-result.csv" [1 2 7] "ant13" "./within") 

(defn data-case [x]
         (case x
           'ant13 1
           'arc 2
           'camel10 3
           'poi15 4
           'redaktor 5
           'skarbonka 6
           'tomcat 7
           'velocity14 8
           'xalan24 9
           'xerces12 10
           (str "else " x)))

(def lr1 (trans [0 0 0 0 0 0]))
(def nb1 (trans [0 0 0 0 0 0]))
(def rf1 (trans [0 0 0 0 0 0]))

(defn extract-performance-data1 [result-csv-name fm test-name test-file-name] ;fm [1 2 7]
	(let [result (report-file result-csv-name)
				result1 (sel result :cols fm)
				gb1 (group-by first result1)
				ml-nb (matrix (second (first (filter #(= 0 (first %)) gb1))))
				ml-lr (matrix (second (first (filter #(= 3 (first %)) gb1))))
				ml-rf (matrix (second (first (filter #(= 4 (first %)) gb1))))
;				ml-knn (matrix (second (first (filter #(= 5 (first %)) gb1))))
				td-nb (group-by #(nth % 1) ml-nb)
				td-lr (group-by #(nth % 1) ml-lr)
				td-rf (group-by #(nth % 1) ml-rf)
;				td-knn (group-by #(nth % 1) ml-knn)
				tbl (fn [group-td-data] 
							(loop [td (sort-by first group-td-data) result []]
								(if (empty? td)
									(matrix result)
									(recur
										(rest td)
;										(conj result (apply vector (cons (first (first td)) (last (trans (second (first td)))))))))))]				
										(conj result (get-median (last (trans (second (first td))))))))))]

			(def mylr (bind-columns mylr (tbl td-lr)))
			(def mynb (bind-columns mylr (tbl td-nb)))
			(def myrf (bind-columns mylr (tbl td-rf)))
				
		;traindata,orig,priv,pcliff1,pcliff2,pcliff4		
		[(save (tbl td-nb) (str test-file-name"/nb-"test-name".dat") :delim \space)
		 (save (tbl td-lr) (str test-file-name"/lr-"test-name".dat") :delim \space)
		 (save (tbl td-rf) (str test-file-name"/rf-"test-name".dat") :delim \space)]
		 
;		[(def lr1 (matrix (data-case test-name) (tbl td-lr)))
;		 (def nb1 (matrix (data-case test-name) (tbl td-nb)))
;		 (def rf1 (matrix (data-case test-name) (tbl td-rf)))]
		 
		 ))
;		 (save (tbl td-knn) (str test-file-name"/knn-"test-name".dat") :delim \space)]))
		 
		 
(defn extract-acc-data2 [result-csv-name a performance-name] ;a for [0 2 3]
	(let [result (report-file result-csv-name) 
				result1 (sel result :cols a)
				;gb0 (group-by first result1)
				ml-rs3 result1
				td-rs3 (group-by #(nth % 1) ml-rs3)
				med-tbl (fn [group-td-data]
							(loop [td (sort-by first group-td-data) result []]
								(if (empty? td)
									result
									(recur
										(rest td)
										(conj result (get-median (last (trans (second (first td))))))))))

     	myans	(matrix [(med-tbl td-rs3)
		                   ])
		    knn-rs3 (sel myans :cols [0 1 2 3 4])
		    cliff-rs3 (sel myans :cols [5 6 7 8 9])
		    ]
		[(save (bind-columns (matrix [1 2 3 4 5]) (trans [knn-rs3])) (str "./code/report/report-2011-11-18-1/"performance-name"-knn.dat") :delim \space)
		(save (bind-columns (matrix [1 2 3 4 5]) (trans [cliff-rs3])) (str "./code/report/report-2011-11-18-1/"performance-name"-cliff.dat") :delim \space)]))
		 
		
		
;for all for rs data

(defn extract-acc-data3 [result-csv-name a performance-name] ;a for [0 2 3]
	(let [result (report-file result-csv-name) 
				result1 (sel result :cols a)
				;gb0 (group-by first result1)
				ml-rs3 result1
				td-rs3 (group-by #(nth % 1) ml-rs3)
				med-tbl (fn [group-td-data]
							(loop [td (sort-by first group-td-data) result []]
								(if (empty? td)
									result
									(recur
										(rest td)
										(conj result (get-median (last (trans (second (first td))))))))))

     	myans	(matrix [(med-tbl td-rs3)
		                   ])]
		                   
		[(sel myans :cols [0 1 2 3 4]) (sel myans :cols [5 6 7 8 9])]))	
		
		                 
;		    knn-rs3 (sel myans :cols [0 1 2 3 4])
;		    cliff-rs3 (sel myans :cols [5 6 7 8 9])
		    
;		[(save (bind-columns (matrix [1 2 3 4 5]) (trans [knn-rs3])) (str "./code/report/report-2011-11-18-1/"performance-name"-knn.dat") :delim \space)
;		(save (bind-columns (matrix [1 2 3 4 5]) (trans [cliff-rs3])) (str "./code/report/report-2011-11-18-1/"performance-name"-cliff.dat") :delim \space)]))	


(defn extract-acc-data4 [result-csv-name a performance-name] ;a for [3 5] 5=pd within
	(let [result (report-file result-csv-name) 
				result1 (sel result :cols a)
				ml-rs3 result1
				td-rs3 (group-by #(nth % 0) ml-rs3)
				med-tbl (fn [group-td-data]
							(loop [td (sort-by first group-td-data) result []]
								(if (empty? td)
									result
									(recur
										(rest td)
										(conj result (get-median (last (trans (second (first td))))))))))
     	myans	(matrix [(med-tbl td-rs3)])]                   
		[(sel myans :cols [0 1 2 3 4 5 6 7])]))



(def rcns [;"./code/report/report-2012-02-01-1/within/rf/ant13-output-result.csv"
					 ;"./code/report/report-2012-02-01-1/within/rf/arc-output-result.csv"
					 ;"./code/report/report-2012-02-01-1/within/rf/camel10-output-result.csv"
					 ;"./code/report/report-2012-02-01-1/within/rf/poi15-output-result.csv"
					 ;"./code/report/report-2012-02-01-1/within/rf/redaktor-output-result.csv"
					 ;"./code/report/report-2012-02-01-1/within/rf/skarbonka-output-result.csv"
					 ;"./code/report/report-2012-02-01-1/within/rf/tomcat-output-result.csv"
					 ;"./code/report/report-2012-02-01-1/within/rf/velocity14-output-result.csv"
					 ;"./code/report/report-2012-02-01-1/within/rf/xalan24-output-result.csv"
					 ;"./code/report/report-2012-02-01-1/within/rf/xerces12-output-result.csv"
					 "./code/report/report-2011-11-18-1/rs3-output-result.csv"
					 "./code/report/report-2011-11-18-1/rs5-output-result.csv"
					 "./code/report/report-2011-11-18-1/rs10-output-result.csv"
					 "./code/report/report-2011-11-18-1/rs20-output-result.csv"])

(defn ead3 [result-csv-names a performance-name]
	(loop [rcn result-csv-names result []]
		(if (empty? rcn)
			(let [;nb-result (trans (map first result))]
						knn-result (trans (map first result))
					  cliff-result (trans (map second result))]
				[(save (bind-columns (matrix [1 2 3 4 5]) knn-result) (str "./code/report/report-2011-11-18-1/"performance-name"-knn.dat") :delim \space)
				 (save (bind-columns (matrix [1 2 3 4 5]) cliff-result) (str "./code/report/report-2011-11-18-1/"performance-name"-cliff.dat") :delim \space)])
			(recur
				(rest rcn)
				(conj result (extract-acc-data3 (first rcn) a performance-name))))))
				
; (ead3 rcns [3 5] 'pd)