Bayes Tests

SS20/Bayes_Tests/test_internal.py

+import sys
+import os
+import importlib.util
+import csv
+
+def makenode(name, parents, *probs):
+    probabilities = {}
+    for i in range(len(probs)//2):
+        combination = tuple({'f' : False, 't' : True}[e] for e in probs[2*i])
+        assert len(combination) == len(parents)
+        assert combination not in probabilities
+        probabilities[combination] = float(probs[2*i+1])
+    assert len(probabilities) == 2**len(parents)
+    return {'name' : name, 'parents' : parents, 'probabilities' : probabilities}
+
+network1 = {
+    'Asia' : makenode('Asia', [], '', 0.05),
+    'Smoke' : makenode('Smoke', [], '', 0.3),
+    'TBC' : makenode('TBC', ['Asia'], 't', 0.01, 'f', 0.001),
+    'LC' : makenode('LC', ['Smoke'], 't', 0.2, 'f', 0.08),
+    'Bron' : makenode('Bron', ['Smoke'], 't', 0.4, 'f', 0.1),
+    'Xray' : makenode('Xray', ['TBC', 'LC'], 'tt', 0.98, 'tf', 0.94, 'ft', 0.92, 'ff', 0.02),
+    'Dysp' : makenode('Dysp', ['TBC', 'LC', 'Bron'],
+            'ttt', 0.99, 'ttf', 0.97, 'tft', 0.98, 'tff', 0.9,
+            'ftt', 0.98, 'ftf', 0.92, 'fft', 0.95, 'fff', 0.07),
+    }
+
+network2 = {
+    'Burglary' : makenode('Burglary', [], '', 0.001),
+    'Earthquake' : makenode('Earthquake', [], '', 0.002),
+    'Alarm' : makenode('Alarm', ['Burglary', 'Earthquake'], 'tt', 0.95, 'tf', 0.94, 'ft', 0.29, 'ff', 0.001),
+    'JohnCalls' : makenode('JohnCalls', ['Alarm'], 't', 0.9, 'f', 0.05),
+    'MaryCalls' : makenode('MaryCalls', ['Alarm'], 't', 0.7, 'f', 0.01),
+    }
+
+network3 = {
+    'Trivial' : makenode('Trivial', [], '', 0.08),
+    }
+
+network4 = {
+    'Deterministic' : makenode('Deterministic', [], '', 1),
+    'Earthquake' : makenode('Earthquake', [], '', 0.002),
+    'Alarm' : makenode('Alarm', ['Deterministic', 'Earthquake'], 'tt', 0.95, 'tf', 0.94, 'ft', 0.29, 'ff', 0.001),
+    'JohnCalls' : makenode('JohnCalls', ['Alarm'], 't', 0.9, 'f', 0.05),
+    'MaryCalls' : makenode('MaryCalls', ['Alarm'], 't', 0.7, 'f', 0.01),
+    }
+
+network5 = {
+    'Burglary' : makenode('Burglary', [], '', 0.001),
+    'Earthquake' : makenode('Earthquake', [], '', 0.002),
+    'Alarm' : makenode('Alarm', ['Burglary', 'Earthquake'], 'tt', 1, 'tf', 0.94, 'ft', 0.29, 'ff', 0.001),
+    'JohnCalls' : makenode('JohnCalls', ['Alarm'], 't', 0.9, 'f', 0.05),
+    'MaryCalls' : makenode('MaryCalls', ['Alarm'], 't', 0.7, 'f', 0.01),
+    }
+
+def give_points(network):
+    if network == network1 :
+        points = 35
+        return points
+    elif network == network2 :
+        points = 35
+        return points
+    elif network == network3 :
+        points = 10
+        return points
+    elif network == network4 :
+        points = 10
+        return points
+    elif network == network5 :
+        points = 10
+        return points
+
+
+def testquery(network, node, evidence, expectedresult):
+     try:
+        estr = ', '.join([e.lower() if evidence[e] else "¬"+e.lower() for e in evidence.keys()])
+        print(f'Query: P({node.lower()} | {estr})')
+        result = bayes.query(network, node, evidence)
+        print(f'Result: {result}')
+        if abs(result - expectedresult) < 1e-10:
+            print('SUCCESS!')
+            points = give_points(network)
+            print("Points: " + str(points))
+            return points
+        else:
+            print(f'I expected {expectedresult}')
+            points = 0
+            print("Points: " + str(points))
+            return points
+     except:
+            e = sys.exc_info()[0]
+            print( "<p>Error: %s</p>" % e )
+            print("Bug in solution! Skipped")
+            points = 0
+            print("Points: " + str(points))
+            return points
+
+
+if __name__ == '__main__':
+    with open(str(os.pardir)+'\\grades.csv', 'a', newline = '') as csvfile:
+        gradeswriter = csv.writer(csvfile, dialect='excel')
+        sys.path.append(os.getcwd())
+        orig_stdout = sys.stdout
+        #Write feedback file:
+        f = open('correction.txt', 'w')
+        sys.stdout = f
+        try:
+            import bayes 
+            points1=testquery(network1, 'TBC', {'Asia':False, 'Xray':True, 'Dysp':True}, 0.008321327685349975)
+            print()
+            points2=testquery(network2, 'Burglary', {'MaryCalls':True, 'JohnCalls':True}, 0.2841718353643929)
+            print()
+            points3=testquery(network3, 'Trivial', {}, 0.08)
+            print()
+            points4=testquery(network4, 'Alarm', {'Deterministic':True, 'Earthquake':True}, 0.9499999999999998)
+            print()
+            points5=testquery(network5, 'Alarm', {'Burglary':True, 'Earthquake':True}, 1)
+            print()
+            finalpoints=points1+points2+points3+points4+points5
+            print("points: "+ str(finalpoints)+"/100")
+            sys.stdout = orig_stdout
+            f.close()
+        # Submissions with non-vanilla dependencies are graded with 0 points.
+        except ModuleNotFoundError:
+            print("Not-included library used, grading not possible.")
+            finalpoints=0 
+        gradeswriter.writerow([str(os.path.basename(os.getcwd())), str(finalpoints)])
+
+