Initial commit — Coursera Big Data coursework

hw7/gene.py

@@ -0,0 +1,17 @@
+import Orange
+
+train = Orange.data.Table("genestrain") #genestrain
+blind = Orange.data.Table("genesblind")
+
+learner = Orange.classification.bayes.NaiveLearner()
+classifier = learner(train)
+classInt = {'CEU' : 0, 'GIH': 1, 'JPT': 2, 'ASW' : 3, 'YRI' : 4 }
+
+results = [str(classifier(d)) for d in blind]
+intResults = [str(classInt[r]) for r in results]
+
+print ' '.join(intResults)
+
+#for d in blind:
+#    c = classifier(d)
+#    print "%10s" % (classifier(d))