score

integr/my_gbdt.py

@@ -18,6 +18,7 @@ def init(Ytrain):
 def fit(Xtrain, Ytrain):
     print("init", Ytrain[:10])
     fx = []
+    trees = []
 
     fx0 = np.ones(Ytrain.shape[0])*init(Ytrain)
     fx.append(fx0)
@@ -26,12 +27,13 @@ def fit(Xtrain, Ytrain):
 
     gx = Ytrain
 
-    for i in range(50):
+    for i in range(200):
         # 求残差
         gx = gx - fx0
         print("第", i, '轮 残差', gx[:10])
-        clf = tree.DecisionTreeRegressor(criterion="mse", max_features=1, max_depth=4)
+        clf = tree.DecisionTreeRegressor(criterion="mse", max_features=1, max_depth=2)
         clf.fit(Xtrain, gx)
+        trees.append(clf)
 
         fx0 = clf.predict(Xtrain)*0.7
         print("第", i, '轮 结果', fx0[:10])
@@ -48,8 +50,26 @@ def fit(Xtrain, Ytrain):
     for i in range(Ytrain.shape[0]):
         sum = sum + (gx[i] - Ytrain[i])*(gx[i] - Ytrain[i])
 
-    print(sum)
+    print("train mse", sum)
+    return trees, fx[0][0]
+
+
+def score(Xtest, Ytest, trees, fx0):
+    gx = np.ones(Ytest.shape[0]) * fx0
+
+    for i in range(len(trees)):
+        gx = gx + trees[i].predict(Xtest)
+    print(gx[:10])
+    print(Ytest[:10])
+
+    sum = 0
+    for i in range(Ytest.shape[0]):
+        sum = sum + (gx[i] - Ytest[i]) * (gx[i] - Ytest[i])
+
+    print("test mse", sum)
+
 
 if __name__ == '__main__':
     Xtrain, Xtest, Ytrain, Ytest = read_data()
-    fit(Xtrain, Ytrain)
+    trees, fx0 = fit(Xtrain, Ytrain)
+    score(Xtest, Ytest, trees, fx0)

integr/my_gbdt_classic.py

@@ -23,9 +23,11 @@ def init(Ytrain):
     p = np.log2(positive/negative)
     return np.ones(Ytrain.shape[0])*p
 
+
 def fit(Xtrain, Ytrain):
     print("init", Ytrain[:10])
     fx = []
+    clf_tress = []
 
 
     fx0 = init(Ytrain)
@@ -35,7 +37,7 @@ def fit(Xtrain, Ytrain):
 
     gx = fx0
 
-    for i in range(50):
+    for i in range(10):
         # 求伪残差
         hx_0 = []
         for j in range(Ytrain.shape[0]):
@@ -43,8 +45,9 @@ def fit(Xtrain, Ytrain):
             hx_0.append(p)
 
         print("第", i, '轮 残差', gx[:10])
-        clf = tree.DecisionTreeRegressor(criterion="mse", max_features=1, max_depth=4)
+        clf = tree.DecisionTreeRegressor(criterion="mse", max_features=1, max_depth=1)
         clf.fit(Xtrain, np.array(hx_0))
+        clf_tress.append(clf)
 
         fx_i = clf.predict(Xtrain)*0.7
         print("第", i, '轮 结果', fx_i[:10])
@@ -61,10 +64,25 @@ def fit(Xtrain, Ytrain):
     gx = np.sign(gx)
     p = sum(gx==Ytrain)/Ytrain.shape[0]
 
-    print(p)
-    return fx
+    print("准确率", p)
+    return clf_tress, fx0[0]
+
+def score(Xtest, Ytest, trees, fx0):
+    gx = np.zeros(Ytest.shape[0])
+
+    for i in range(len(trees)):
+        gx = gx + trees[i].predict(Xtest)
+
+    gx = np.sign(gx)
+    p = sum(gx == Ytest) / Ytest.shape[0]
+    print("准确率", p)
+
+    gx = np.sign(trees[0].predict(Xtest))
+    p = sum(gx == Ytest) / Ytest.shape[0]
+    print("准确率0", p)
 
 
 if __name__ == '__main__':
     Xtrain, Xtest, Ytrain, Ytest = read_data()
-    fx = fit(Xtrain, Ytrain)
+    trees,fx0 = fit(Xtrain, Ytrain)
+    score(Xtest, Ytest, trees, fx0)