update

Cart.py

@@ -18,31 +18,27 @@ def __init__(self):
     #type : 0 - list, 1 - node
 
     def sum(self, X, p):
-        s = 0
-        for i in range(0, len(X)):
-            s += float(X[i][p])
-        return s
+        return sum(map(float, X[:][p]))
 
     def create_list(self, X, y):
         self.type = 0
         self.attribute = [0, 0, 0, 0]
         self.answer = sum(y) / float(len(X))
 
-
     def is_list_criteria(self, depth):
-        if depth < 3: # Maximum DEPTH
-            return 0
-        else:
-            return 1
+        return depth >= 3 # Maximal DEPTH
 
+    def get_params(self, deep = True):
+        return {}
 
-    #We divide all members into two classes
-    #In first current attribute i < average of all, in second >=
+    # We divide all members into two classes.
+    # In the first one a current attribute _i_ < average of all, in the second >=.
     @staticmethod
     def gini_impurity(self, X, y, depth):
         if self.is_list_criteria(depth):
             self.create_list(X, y)
             return self
+
         self.type = 1
         self.attribute[3] = -1
         for i in range(0, len(X[0])):
@@ -75,17 +71,8 @@ def gini_impurity(self, X, y, depth):
             self.create_list(X, y)
         return self
 
-    def build_tree(self, X, y):
-        self = self.gini_impurity(self, X, y, 0)
-        return self
-
     def fit(self, X, y):
-        t = self.build_tree(X, y)
-        self.attribute = t.attribute
-        self.answer = t.answer
-        self.type = t.type
-        self.left = t.left
-        self.right = t.right
+        return self.gini_impurity(self, X, y, 0)
 
     def predict_user(self, user):
         if self.type:
@@ -97,13 +84,7 @@ def predict_user(self, user):
             return self.answer
 
     def predict(self, X):
-        res = []
-        for i in range(0, len(X)):
-            res.append(self.predict_user(X[i]))
-        return res
-
-    def get_params(self, deep = True):
-        return {}
+        return list(map(self.predict_user, X))
 
     def score(self, X, y):
         y_prediction = numpy.array(self.predict(X)) - numpy.array(y)
@@ -147,7 +128,7 @@ def launch(flearn, ftest):
             tk = line.split(',')
             t = [float(tk[3]), float(tk[4]), float(tk[6]), float(tk[7]), float(tk[8])]
             temp = CART.predict_user(t)
-            print(str([tk + ' prediction is ' + temp]))
+            print(''.join(map(str, [tk, ' prediction is ', temp])))
             sumsq += (float(tk[-1]) - temp) ** 2
             i += 1
             line = f.readline()
@@ -156,7 +137,7 @@ def launch(flearn, ftest):
             tk = tk[2].split(',')
             t = [float(tk[1]), float(tk[2]), float(tk[4]), float(tk[5]), float(tk[6])]
             temp = CART.predict_user(t)
-            print(str([tk + ' prediction is ' + temp]))
+            print(''.join(map(str, [tk, ' prediction is ', temp])))
             sumsq += (float(tk[-1]) - temp) ** 2
             i += 1
             line = f.readline()

README.md

@@ -1,11 +1,9 @@
-Decision_tree-CART-
-===================
+### Decision tree impelentation, *CART*
 
-Homework for Data Mining
+*   learn.csv - train set
+*   test.csv - test set
+*   users.csv - the full data
 
-learn.csv - обучающая выборка test.csv - тестовая выборка
+Run and test: `python3 Cart.py learn.csv users.csv`
 
-users.csv - общий набор данных
-
-обучить и протестировать: python Cart.py learn.csv userfilt.csv 
-кросс-валидация: python Cart.py cross-val users.csv
+Cross-validate: `python3 Cart.py cross-val users.csv`