new

PENGZhaoqing · Dec 1, 2016 · 78a3e70 · 78a3e70
1 parent 7c980d8
commit 78a3e70
Show file tree

Hide file tree

Showing 9 changed files with 50,007 additions and 41,775 deletions.
diff --git a/CART/CART_Trainer.py b/CART/CART_Trainer.py
@@ -16,7 +16,7 @@ class Trainer:
 
     def train(self, training_set, training_target, fea_index):
 
-        clf = tree.DecisionTreeClassifier(criterion="gini", max_depth=7, min_samples_leaf=50)
+        clf = tree.DecisionTreeClassifier(criterion="entropy", min_samples_split=30, class_weight="balanced")
         clf = clf.fit(training_set, training_target)
 
         class_names = np.unique([str(i) for i in training_target])

diff --git a/CART/CART_Trainer.pyc b/CART/CART_Trainer.pyc
diff --git a/CART/output/.DS_Store b/CART/output/.DS_Store
diff --git a/CART/output/CART.pkl b/CART/output/CART.pkl
diff --git a/CART/output/trained_text.txt b/CART/output/trained_text.txt
diff --git a/CART/output/tree-vis.pdf b/CART/output/tree-vis.pdf
diff --git a/Mongo_Con.py b/Mongo_Con.py
@@ -44,11 +44,11 @@ def import_test_data(self, test_file):
                 self.db.test_data.insert_one({"test_set": dic})
 
     def CART_fetch_data(self):
-        training_cursor = self.db.training_data.find({"training_set.src_bytes": {"$gt": 1000}})
-        test_cursor = self.db.test_data.find({"test_set.src_bytes": {"$gt": 1000}})
+        # training_cursor = self.db.training_data.find({"training_set.src_bytes": {"$gt": 1000}})
+        # test_cursor = self.db.test_data.find({"test_set.src_bytes": {"$gt": 1000}})
 
-        # training_cursor = self.db.training_data.find()
-        # test_cursor = self.db.test_data.find()
+        training_cursor = self.db.training_data.find()
+        test_cursor = self.db.test_data.find()
 
         cursor = training_cursor.sort('training_set.type', pymongo.ASCENDING)
 

diff --git a/Mongo_Con.pyc b/Mongo_Con.pyc
diff --git a/Preprocessing_all.py b/Preprocessing_all.py
@@ -144,3 +144,5 @@ def isfloat(value):
                 element = float(element)
             dic[attr] = element
         db.test_data.insert_one({"test_set": dic})
+
+
Original file line number	Diff line number	Diff line change
Expand Up		@@ -144,3 +144,5 @@ def isfloat(value):
		element = float(element)
		dic[attr] = element
		db.test_data.insert_one({"test_set": dic})