Lisp version fully complete now.

lisp/example.lisp

@@ -0,0 +1,164 @@
+(defun prune (x)
+  (cond ((null x) nil)
+        ((null (car x)) (prune (cdr x)))
+        (T (cons (car x) (prune (cdr x))))))
+
+(defun read-exampleset (file)
+   (LET ((STREAM (OPEN file :DIRECTION :INPUT)))
+     (DO ((expression NIL (READ STREAM NIL STREAM)) 
+          (xprlist nil (cons expression xprlist)))
+         ((EQ expression STREAM) (progn (CLOSE STREAM) (reverse (prune xprlist)))))))
+
+; --- example:
+
+; Generate a file in which you may write the lecture example:
+; 
+; ("saengerin" "jazz" "20er-50er")
+; ("saenger" "jazz" "20er-50er")
+;
+; (read-exampleset "yourfilename") should then return the list
+;
+; (("saengerin" "jazz" "20er-50er")("saenger" "jazz" "20er-50er"))
+;
+
+(load "version_space.lisp")
+
+(defun negative-examples (examples key)
+  (remove-if #'(lambda (example)
+    (equal (first (last example)) key)
+  ) examples)
+)
+
+(defun positive-examples (examples key)
+  (remove-if-not #'(lambda (example)
+    (equal (first (last example)) key)
+  ) examples)
+)
+
+(defun gen-star (example examples)
+  (first
+    (reduce #'(lambda (vs neg_example)
+      (negative-example vs (butlast neg_example))
+    ) (negative-examples examples (last example)) :initial-value (positive-example (gen-version-space (- (length example) 1)) (butlast example)))
+  )
+)
+
+(defun select-best-generalization (star positives negatives)
+  ;; "Beste" Generalisierung - Erste Generalisierung
+  ; (first star)
+
+  ;; "Beste" Generalisierung - Letzte Generalisierung
+  (first (last star))
+
+  ;; "Beste" Generalisierung - Viele "*"
+  ; (first (sort star #'> :key #'(lambda (g)
+  ;   (reduce (lambda (acc item)
+  ;     (if (equal item "*") (+ acc 1) acc)
+  ;   ) g :initial-value 0)
+  ; )))
+
+  ;; "Beste" Generalisierung - Wenig "*"
+  ; (first (sort star #'> :key #'(lambda (g)
+  ;   (reduce (lambda (acc item)
+  ;     (if (equal item "*") (+ acc 1) acc)
+  ;   ) g :initial-value 0)
+  ; )))
+
+  ;; "Beste" Generalisierung - Möglichst viele Positive Beispiele abgedeckt
+  ; (first (sort star #'> :key #'(lambda (g)
+  ;   (reduce (lambda (acc item)
+  ;     (if (more-general? g item) (+ acc 1) acc)
+  ;   ) positives :initial-value 0)
+  ; )))
+)
+
+(defun gen-best-generalization (pos_example negatives positives)
+  (select-best-generalization
+    (gen-star pos_example negatives)
+    positives
+    negatives
+  )
+)
+
+(defun gen-concept-space-helper (positives negatives k)
+  (if
+    (equal 0 (length positives)) k
+    (let ((g (gen-best-generalization (first positives) negatives positives)))
+      (gen-concept-space-helper
+        (remove-if #'(lambda (pos_example)
+          (more-general? g pos_example)
+        ) positives)
+      negatives (cons g k))
+    )
+  )
+)
+
+(defun gen-concept-space (positives negatives)
+  (gen-concept-space-helper positives negatives '())
+)
+
+(defun matches-for (example K)
+  (reduce #'(lambda (result concept_space)
+    (append result (reduce #'(lambda (matches generalization)
+      (if (more-general? generalization (butlast example))
+        (append matches (list (first concept_space)))
+        matches
+      )
+    ) (second concept_space) :initial-value '()) )
+  ) K :initial-value '())
+)
+
+(let
+  ( (examples (read-exampleset "input.txt"))
+    (test_examples (read-exampleset "test.txt"))
+    (keys '("Buch_A" "Buch_B" "Buch_C"))
+  )
+
+  (let (
+    (K (mapcar #'(lambda (key)
+      (list key
+        (gen-concept-space
+          (positive-examples examples key)
+          (negative-examples examples key)
+        )
+      )
+    ) keys)))
+
+    (let (
+      (matches (mapcar #'(lambda (example)
+        (list example (matches-for example K))
+      ) test_examples)))
+
+      (loop for match in matches do
+        (format t "---------------------------------------~%")
+        (format t "Datensatz: ~A~%" (butlast (first match)))
+        (if (second match)
+          (format t "Mögliche Buchwahl: ~A~%" (remove-duplicates (second match)))
+          (format t "Leider keine Aussage möglich!~%")
+        )
+        (format t "Die korrekte Buchwahl: ~A" (last (first match)))
+        (format t "~%")
+        (format t "~%")
+      )
+
+      (format t "---------------------------------------~%")
+
+      (format t "Anzahl an korrekten Vorhersagen ~A~%" (count-if #'(lambda (match)
+        (and 
+          (equal (list-length (second match)) 1)
+          (find-if #'(lambda (x) (equal x (first (second match)))) (second match))
+        )
+      ) matches))
+
+      (format t "Anzahl an nicht eindeutigen Vorhersagen ~A~%" (count-if #'(lambda (match)
+        (and 
+          (> (list-length (second match)) 1)
+        )
+      ) matches))
+
+      (format t "Anzahl an falschen Vorhersagen ~A~%" (count-if #'(lambda (match)
+        (not (find-if #'(lambda (x) (equal x (first (second match)))) (second match)))
+      ) matches))
+    )
+  )
+)

lisp/input.txt

@@ -0,0 +1,70 @@
+("19-24" "m" "nein" "0" "keiner" "Angestellter" "3000-3999" "Buch_C")
+("25-35" "m" "nein" "1" "Realschule" "Arbeitslos" "1000-1999" "Buch_C")
+("36-49" "m" "nein" "0" "Hochschule" "Angestellter" "3000-3999" "Buch_C")
+("25-35" "m" "ja" "1" "Hochschule" "Fuehrungskraft" "5000 und mehr" "Buch_C")
+("36-49" "w" "nein" "0" "Gymnasium" "Angestellter" "2000-2999" "Buch_C")
+("<18" "w" "nein" "0" "keiner" "Hausfrau" "2000-2999" "Buch_A")
+("19-24" "w" "ja" "1" "Hauptschule" "Arbeitslos" "1000-1999" "Buch_B")
+("25-35" "w" "ja" "1" "Hochschule" "Arbeitslos" "<1000" "Buch_C")
+("<18" "w" "nein" "0" "keiner" "Angestellter" "2000-2999" "Buch_A")
+("19-24" "w" "nein" "0" "keiner" "Selbstaendig" "1000-1999" "Buch_A")
+("50-65" "m" "nein" "0" "Promotion" "Fuehrungskraft" "5000 und mehr" "Buch_C")
+("<18" "m" "nein" "0" "Realschule" "Angestellter" "3000-3999" "Buch_C")
+("<18" "w" "ja" "1" "keiner" "Arbeitslos" "<1000" "Buch_A")
+("36-49" "m" "nein" "0" "Hochschule" "Angestellter" "2000-2999" "Buch_C")
+("36-49" "w" "ja" "2" "Hochschule" "Arbeiter" "1000-1999" "Buch_C")
+("36-49" "w" "ja" "0" "Realschule" "Angestellter" "1000-1999" "Buch_C")
+("36-49" "w" "nein" "0" "Hochschule" "Hausfrau" "3000-3999" "Buch_C")
+("25-35" "m" "nein" "1" "Hochschule" "Angestellter" "3000-3999" "Buch_C")
+("50-65" "m" "nein" "0" "keiner" "Rentner" "1000-1999" "Buch_A")
+("50-65" "w" "nein" "1" "Hochschule" "Angestellter" "2000-2999" "Buch_C")
+("25-35" "m" "nein" "1" "Gymnasium" "Selbstaendig" "5000 und mehr" "Buch_C")
+("19-24" "m" "ja" "0" "Gymnasium" "Angestellter" "2000-2999" "Buch_C")
+("25-35" "w" "ja" "4" "Hochschule" "Hausfrau" "2000-2999" "Buch_C")
+("36-49" "w" "ja" "1" "keiner" "Angestellter" "2000-2999" "Buch_A")
+("50-65" "m" "ja" "2" "Hauptschule" "Angestellter" "2000-2999" "Buch_C")
+("19-24" "m" "nein" "0" "Gymnasium" "Angestellter" "3000-3999" "Buch_C")
+("36-49" "m" "ja" "0" "Hochschule" "Angestellter" "2000-2999" "Buch_C")
+("36-49" "w" "ja" "2" "keiner" "Arbeiter" "1000-1999" "Buch_A")
+("25-35" "m" "nein" "0" "Hochschule" "Arbeitslos" "<1000" "Buch_C")
+("50-65" "w" "nein" "0" "Promotion" "Hausfrau" "3000-3999" "Buch_C")
+("<18" "w" "nein" "0" "keiner" "Hausfrau" "2000-2999" "Buch_A")
+("36-49" "m" "ja" "2" "Hochschule" "Selbstaendig" "2000-2999" "Buch_C")
+("<18" "m" "nein" "0" "Realschule" "Angestellter" "1000-1999" "Buch_C")
+("36-49" "m" "nein" "0" "keiner" "Angestellter" "1000-1999" "Buch_A")
+("36-49" "w" "nein" "0" "Hochschule" "Selbstaendig" "5000 und mehr" "Buch_C")
+("19-24" "w" "nein" "0" "keiner" "Selbstaendig" "3000-3999" "Buch_B")
+("50-65" "m" "nein" "0" "Hochschule" "Rentner" "2000-2999" "Buch_C")
+("<18" "m" "nein" "0" "keiner" "Angestellter" "2000-2999" "Buch_A")
+("19-24" "w" "nein" "0" "keiner" "Hausfrau" "4000-4999" "Buch_B")
+("25-35" "m" "nein" "0" "keiner" "Arbeiter" "2000-2999" "Buch_A")
+("25-35" "w" "nein" "0" "Hochschule" "Selbstaendig" "2000-2999" "Buch_C")
+("25-35" "w" "nein" "1" "keiner" "Hausfrau" "4000-4999" "Buch_C")
+("25-35" "w" "nein" "0" "Realschule" "Angestellter" "2000-2999" "Buch_C")
+("36-49" "m" "ja" "1" "Hochschule" "Fuehrungskraft" "5000 und mehr" "Buch_C")
+("36-49" "m" "nein" "1" "Hauptschule" "Angestellter" "3000-3999" "Buch_C")
+("25-35" "m" "nein" "0" "Hochschule" "Selbstaendig" "3000-3999" "Buch_C")
+("25-35" "m" "nein" "1" "Hochschule" "Lehrer" "2000-2999" "Buch_C")
+("19-24" "w" "nein" "0" "Hauptschule" "Angestellter" "2000-2999" "Buch_B")
+("25-35" "w" "ja" "1" "Hochschule" "Lehrer" "2000-2999" "Buch_C")
+("25-35" "m" "ja" "3" "Promotion" "Fuehrungskraft" "5000 und mehr" "Buch_C")
+("25-35" "w" "ja" "0" "Hochschule" "Hausfrau" "3000-3999" "Buch_C")
+("19-24" "w" "nein" "0" "keiner" "Angestellter" "3000-3999" "Buch_B")
+("<18" "m" "ja" "0" "keiner" "Angestellter" "2000-2999" "Buch_A")
+("36-49" "w" "ja" "3" "keiner" "Angestellter" "2000-2999" "Buch_A")
+("<18" "w" "ja" "0" "Realschule" "Angestellter" "2000-2999" "Buch_B")
+("19-24" "m" "nein" "0" "keiner" "Arbeiter" "1000-1999" "Buch_A")
+("<18" "m" "nein" "0" "keiner" "Angestellter" "3000-3999" "Buch_C")
+("36-49" "m" "nein" "1" "Hauptschule" "Angestellter" "2000-2999" "Buch_C")
+("19-24" "w" "nein" "0" "keiner" "Hausfrau" "2000-2999" "Buch_A")
+("19-24" "w" "nein" "0" "Hauptschule" "Hausfrau" "1000-1999" "Buch_B")
+("19-24" "w" "nein" "0" "Gymnasium" "Angestellter" "2000-2999" "Buch_B")
+("19-24" "w" "nein" "0" "Gymnasium" "Angestellter" "2000-2999" "Buch_B")
+(">65" "w" "ja" "1" "Hochschule" "Rentner" "2000-2999" "Buch_C")
+("36-49" "w" "nein" "0" "keiner" "Hausfrau" "<1000" "Buch_A")
+("36-49" "w" "nein" "0" "Gymnasium" "Angestellter" "2000-2999" "Buch_C")
+("19-24" "m" "nein" "0" "Realschule" "Arbeitslos" "2000-2999" "Buch_C")
+("25-35" "m" "nein" "1" "keiner" "Arbeitslos" "3000-3999" "Buch_C")
+("50-65" "m" "ja" "2" "Realschule" "Arbeiter" "2000-2999" "Buch_C")
+("19-24" "w" "ja" "0" "Gymnasium" "Hausfrau" "3000-3999" "Buch_B")
+("36-49" "w" "ja" "0" "Hochschule" "Angestellter" "2000-2999" "Buch_C")

lisp/test.txt

@@ -0,0 +1,100 @@
+("19-24" "m" "nein" "0" "keiner" "Angestellter" "3000-3999" "Buch_C")
+("25-35" "m" "nein" "1" "Realschule" "Arbeitslos" "1000-1999" "Buch_C")
+("36-49" "m" "nein" "0" "Hochschule" "Angestellter" "3000-3999" "Buch_C")
+("25-35" "m" "ja" "1" "Hochschule" "Fuehrungskraft" "5000 und mehr" "Buch_C")
+("36-49" "w" "nein" "0" "Gymnasium" "Angestellter" "2000-2999" "Buch_C")
+("<18" "w" "nein" "0" "keiner" "Hausfrau" "2000-2999" "Buch_A")
+("19-24" "w" "ja" "1" "Hauptschule" "Arbeitslos" "1000-1999" "Buch_B")
+("25-35" "w" "ja" "1" "Hochschule" "Arbeitslos" "<1000" "Buch_C")
+("<18" "w" "nein" "0" "keiner" "Angestellter" "2000-2999" "Buch_A")
+("19-24" "w" "nein" "0" "keiner" "Selbstaendig" "1000-1999" "Buch_A")
+("50-65" "m" "nein" "0" "Promotion" "Fuehrungskraft" "5000 und mehr" "Buch_C")
+("<18" "m" "nein" "0" "Realschule" "Angestellter" "3000-3999" "Buch_C")
+("<18" "w" "ja" "1" "keiner" "Arbeitslos" "<1000" "Buch_A")
+("36-49" "m" "nein" "0" "Hochschule" "Angestellter" "2000-2999" "Buch_C")
+("36-49" "w" "ja" "2" "Hochschule" "Arbeiter" "1000-1999" "Buch_C")
+("36-49" "w" "ja" "0" "Realschule" "Angestellter" "1000-1999" "Buch_C")
+("36-49" "w" "nein" "0" "Hochschule" "Hausfrau" "3000-3999" "Buch_C")
+("25-35" "m" "nein" "1" "Hochschule" "Angestellter" "3000-3999" "Buch_C")
+("50-65" "m" "nein" "0" "keiner" "Rentner" "1000-1999" "Buch_A")
+("50-65" "w" "nein" "1" "Hochschule" "Angestellter" "2000-2999" "Buch_C")
+("25-35" "m" "nein" "1" "Gymnasium" "Selbstaendig" "5000 und mehr" "Buch_C")
+("19-24" "m" "ja" "0" "Gymnasium" "Angestellter" "2000-2999" "Buch_C")
+("25-35" "w" "ja" "4" "Hochschule" "Hausfrau" "2000-2999" "Buch_C")
+("36-49" "w" "ja" "1" "keiner" "Angestellter" "2000-2999" "Buch_A")
+("50-65" "m" "ja" "2" "Hauptschule" "Angestellter" "2000-2999" "Buch_C")
+("19-24" "m" "nein" "0" "Gymnasium" "Angestellter" "3000-3999" "Buch_C")
+("36-49" "m" "ja" "0" "Hochschule" "Angestellter" "2000-2999" "Buch_C")
+("36-49" "w" "ja" "2" "keiner" "Arbeiter" "1000-1999" "Buch_A")
+("25-35" "m" "nein" "0" "Hochschule" "Arbeitslos" "<1000" "Buch_C")
+("50-65" "w" "nein" "0" "Promotion" "Hausfrau" "3000-3999" "Buch_C")
+("<18" "w" "nein" "0" "keiner" "Hausfrau" "2000-2999" "Buch_A")
+("36-49" "m" "ja" "2" "Hochschule" "Selbstaendig" "2000-2999" "Buch_C")
+("<18" "m" "nein" "0" "Realschule" "Angestellter" "1000-1999" "Buch_C")
+("36-49" "m" "nein" "0" "keiner" "Angestellter" "1000-1999" "Buch_A")
+("36-49" "w" "nein" "0" "Hochschule" "Selbstaendig" "5000 und mehr" "Buch_C")
+("19-24" "w" "nein" "0" "keiner" "Selbstaendig" "3000-3999" "Buch_B")
+("50-65" "m" "nein" "0" "Hochschule" "Rentner" "2000-2999" "Buch_C")
+("<18" "m" "nein" "0" "keiner" "Angestellter" "2000-2999" "Buch_A")
+("19-24" "w" "nein" "0" "keiner" "Hausfrau" "4000-4999" "Buch_B")
+("25-35" "m" "nein" "0" "keiner" "Arbeiter" "2000-2999" "Buch_A")
+("25-35" "w" "nein" "0" "Hochschule" "Selbstaendig" "2000-2999" "Buch_C")
+("25-35" "w" "nein" "1" "keiner" "Hausfrau" "4000-4999" "Buch_C")
+("25-35" "w" "nein" "0" "Realschule" "Angestellter" "2000-2999" "Buch_C")
+("36-49" "m" "ja" "1" "Hochschule" "Fuehrungskraft" "5000 und mehr" "Buch_C")
+("36-49" "m" "nein" "1" "Hauptschule" "Angestellter" "3000-3999" "Buch_C")
+("25-35" "m" "nein" "0" "Hochschule" "Selbstaendig" "3000-3999" "Buch_C")
+("25-35" "m" "nein" "1" "Hochschule" "Lehrer" "2000-2999" "Buch_C")
+("19-24" "w" "nein" "0" "Hauptschule" "Angestellter" "2000-2999" "Buch_B")
+("25-35" "w" "ja" "1" "Hochschule" "Lehrer" "2000-2999" "Buch_C")
+("25-35" "m" "ja" "3" "Promotion" "Fuehrungskraft" "5000 und mehr" "Buch_C")
+("25-35" "w" "ja" "0" "Hochschule" "Hausfrau" "3000-3999" "Buch_C")
+("19-24" "w" "nein" "0" "keiner" "Angestellter" "3000-3999" "Buch_B")
+("<18" "m" "ja" "0" "keiner" "Angestellter" "2000-2999" "Buch_A")
+("36-49" "w" "ja" "3" "keiner" "Angestellter" "2000-2999" "Buch_A")
+("<18" "w" "ja" "0" "Realschule" "Angestellter" "2000-2999" "Buch_B")
+("19-24" "m" "nein" "0" "keiner" "Arbeiter" "1000-1999" "Buch_A")
+("<18" "m" "nein" "0" "keiner" "Angestellter" "3000-3999" "Buch_C")
+("36-49" "m" "nein" "1" "Hauptschule" "Angestellter" "2000-2999" "Buch_C")
+("19-24" "w" "nein" "0" "keiner" "Hausfrau" "2000-2999" "Buch_A")
+("19-24" "w" "nein" "0" "Hauptschule" "Hausfrau" "1000-1999" "Buch_B")
+("19-24" "w" "nein" "0" "Gymnasium" "Angestellter" "2000-2999" "Buch_B")
+("19-24" "w" "nein" "0" "Gymnasium" "Angestellter" "2000-2999" "Buch_B")
+(">65" "w" "ja" "1" "Hochschule" "Rentner" "2000-2999" "Buch_C")
+("36-49" "w" "nein" "0" "keiner" "Hausfrau" "<1000" "Buch_A")
+("36-49" "w" "nein" "0" "Gymnasium" "Angestellter" "2000-2999" "Buch_C")
+("19-24" "m" "nein" "0" "Realschule" "Arbeitslos" "2000-2999" "Buch_C")
+("25-35" "m" "nein" "1" "keiner" "Arbeitslos" "3000-3999" "Buch_C")
+("50-65" "m" "ja" "2" "Realschule" "Arbeiter" "2000-2999" "Buch_C")
+("19-24" "w" "ja" "0" "Gymnasium" "Hausfrau" "3000-3999" "Buch_B")
+("36-49" "w" "ja" "0" "Hochschule" "Angestellter" "2000-2999" "Buch_C")
+("50-65" "w" "nein" "0" "Realschule" "Angestellter" "1000-1999" "Buch_C")
+("50-65" "m" "nein" "0" "Realschule" "Arbeitslos" "1000-1999" "Buch_C")
+("36-49" "m" "nein" "0" "Realschule" "Angestellter" "2000-2999" "Buch_C")
+("19-24" "m" "nein" "0" "keiner" "Arbeitslos" "1000-1999" "Buch_A")
+("<18" "m" "nein" "0" "keiner" "Angestellter" "2000-2999" "Buch_A")
+("36-49" "m" "ja" "3" "keiner" "Arbeiter" "2000-2999" "Buch_A")
+("19-24" "m" "nein" "0" "Realschule" "Arbeiter" "2000-2999" "Buch_C")
+("19-24" "m" "nein" "0" "keiner" "Arbeiter" "3000-3999" "Buch_C")
+("36-49" "m" "nein" "0" "Realschule" "Arbeitslos" "1000-1999" "Buch_C")
+("19-24" "w" "ja" "1" "Gymnasium" "Hausfrau" "2000-2999" "Buch_B")
+("19-24" "m" "nein" "0" "keiner" "Angestellter" "2000-2999" "Buch_A")
+("25-35" "m" "nein" "0" "keiner" "Arbeiter" "1000-1999" "Buch_A")
+("25-35" "m" "nein" "1" "Hochschule" "Angestellter" "2000-2999" "Buch_C")
+("36-49" "m" "nein" "1" "Hochschule" "Fuehrungskraft" "5000 und mehr" "Buch_C")
+("19-24" "w" "nein" "0" "keiner" "Angestellter" "1000-1999" "Buch_A")
+("19-24" "m" "nein" "0" "keiner" "Arbeiter" "2000-2999" "Buch_A")
+("36-49" "m" "nein" "0" "keiner" "Arbeitslos" "1000-1999" "Buch_A")
+("<18" "w" "nein" "0" "keiner" "Angestellter" "2000-2999" "Buch_A")
+("36-49" "m" "ja" "3" "keiner" "Selbstaendig" "1000-1999" "Buch_A")
+("36-49" "m" "nein" "0" "keiner" "Angestellter" "3000-3999" "Buch_C")
+("19-24" "m" "nein" "0" "Gymnasium" "Arbeiter" "2000-2999" "Buch_C")
+("19-24" "w" "nein" "0" "keiner" "Angestellter" "2000-2999" "Buch_A")
+("<18" "w" "nein" "0" "keiner" "Arbeiter" "1000-1999" "Buch_A")
+("<18" "m" "nein" "0" "Realschule" "Arbeiter" "2000-2999" "Buch_C")
+("19-24" "m" "nein" "0" "keiner" "Arbeiter" "2000-2999" "Buch_A")
+("25-35" "w" "nein" "0" "Realschule" "Angestellter" "2000-2999" "Buch_C")
+("19-24" "m" "nein" "1" "Realschule" "Arbeitslos" "1000-1999" "Buch_C")
+("50-65" "m" "ja" "1" "keiner" "Arbeiter" "2000-2999" "Buch_A")
+("19-24" "m" "nein" "0" "Hochschule" "Angestellter" "3000-3999" "Buch_C")
+(">65" "m" "nein" "0" "Realschule" "Rentner" "1000-1999" "Buch_C")