1. tensorflow-iris-example.git
  2. tutorial.py
Browse code

버그 수정

Ikseon Kang authored on29/06/2019 13:26:01
Showing1 changed files
tutorial.py
History View file @0fbd020
... ... 
@@ -4,8 +4,8 @@ import iris_data
4 4
5 5 
 (train_x, train_y), (test_x, test_y) = iris_data.load_data()
6 6
7 
-print np.shape(train_x)
8 
-print np.shape(train_y)
7 
+print (np.shape(train_x))
8 
+print (np.shape(train_y))
9 9
10 10 
 feature_columns = []
11 11 
 for key in train_x.keys():
... ... 
@@ -41,7 +41,7 @@ def test_input_fn(x, y, batch_size):
41 41 
     return dataset
42 42
43 43 
 result = classifier.evaluate(input_fn = lambda:iris_data.eval_input_fn(test_x, test_y, batch_size))
44 
-print result
44 
+print (result)
45 45
46 46 
 predict_x = {
47 47 
     'SepalLength': [5.1, 5.9, 6.9],
... ... 
@@ -53,4 +53,4 @@ predict_x = {
53 53 
 predictions = classifier.predict(input_fn=lambda:test_input_fn(predict_x, y=None, batch_size=batch_size))
54 54
55 55 
 for p in predictions:
56 
-    print p
56 
+    print (p)
Browse code

실행 파일 추가

Nepirity Corp authored on17/04/2018 15:34:56
Showing1 changed files
tutorial.py
History View file @8d83cd4
1 1 
new file mode 100644
... ... 
@@ -0,0 +1,56 @@
1 
+import tensorflow as tf
2 
+import numpy as np
3 
+import iris_data
4 
+
5 
+(train_x, train_y), (test_x, test_y) = iris_data.load_data()
6 
+
7 
+print np.shape(train_x)
8 
+print np.shape(train_y)
9 
+
10 
+feature_columns = []
11 
+for key in train_x.keys():
12 
+    feature_columns.append(tf.feature_column.numeric_column(key))
13 
+
14 
+classifier = tf.estimator.DNNClassifier(
15 
+    feature_columns = feature_columns,
16 
+    hidden_units=[10, 10],
17 
+    n_classes = 3,
18 
+)
19 
+
20 
+batch_size = 100
21 
+steps = 1000
22 
+
23 
+def train_input_fn(x, y, batch_size):
24 
+    dataset = tf.data.Dataset.from_tensor_slices((dict(x), y))
25 
+    dataset = dataset.shuffle(1000).repeat().batch(batch_size)
26 
+
27 
+    return dataset
28 
+
29 
+classifier.train(input_fn = lambda:train_input_fn(train_x, train_y, batch_size), steps = steps)
30 
+
31 
+def test_input_fn(x, y, batch_size):
32 
+    x=dict(x)
33 
+    if y is None:
34 
+        inputs = x
35 
+    else:
36 
+        inputs = (x, y)
37 
+
38 
+    dataset = tf.data.Dataset.from_tensor_slices(inputs)
39 
+    dataset = dataset.batch(batch_size)
40 
+
41 
+    return dataset
42 
+
43 
+result = classifier.evaluate(input_fn = lambda:iris_data.eval_input_fn(test_x, test_y, batch_size))
44 
+print result
45 
+
46 
+predict_x = {
47 
+    'SepalLength': [5.1, 5.9, 6.9],
48 
+    'SepalWidth': [3.3, 3.0, 3.1],
49 
+    'PetalLength': [1.7, 4.2, 5.4],
50 
+    'PetalWidth': [0.5, 1.5, 2.1],
51 
+}
52 
+
53 
+predictions = classifier.predict(input_fn=lambda:test_input_fn(predict_x, y=None, batch_size=batch_size))
54 
+
55 
+for p in predictions:
56 
+    print p