Question

我在tensorflow上使用logistic regression后，尝试获得测试数据的准确性。我用

# parameters
learning_rate = 0.01
training_epochs = 5
batch_size = 8192
display_step = 1

reset_graph()

X = tf.placeholder(tf.float32, shape=(None, X_train.shape[1]), name="X")
y = tf.placeholder(tf.float32, shape=(None, y_train.shape[1]), name="y")

# set model weights
weights = tf.Variable(tf.random_normal([X_train.shape[1], y_train.shape[1]], stddev=0.1), name="weights")
biases = tf.Variable(tf.ones(shape=y_train.shape[1]), name="biases")

# construct model
y_pred = tf.nn.sigmoid(tf.add(tf.matmul(X, weights), biases))

# minimize error using cross entropy
cost = tf.reduce_mean(-(y*tf.log(y_pred) + (1 - y)*tf.log(1 - y_pred)))

# gradient descent
optimizer = tf.train.GradientDescentOptimizer(learning_rate).minimize(cost)

# check accuracy
correct_prediction = tf.equal(y, y_pred)
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))

# variables initializing
init = tf.global_variables_initializer()

# starting session
with tf.Session() as sess:
    sess.run(init)

    for epoch in range(training_epochs):
        for batch_i, (start, end) in enumerate(split(0, X_train.shape[0], batch_size)):
            x_batch, y_true_batch, = X_train[start:end].toarray(), y_train[start:end]
            feed_dict_train = {X: x_batch, y:y_true_batch}
            sess.run(optimizer, feed_dict=feed_dict_train)


        # Display logs per epoch step
        if (epoch + 1) % display_step == 0:
            cost_step, acc_step = sess.run([cost, accuracy], feed_dict={X: X_test.toarray(), y: y_test})

            print("Epoch:", '%04d' % (epoch + 1), "cost =", "{:.4f}".format(cost_step), "accuracy =", "{:.4f}".format(acc_step))
            print(accuracy.eval(feed_dict={X: X_test.toarray(), y: y_test}))

            predicts = sess.run(y_pred, feed_dict={X: X_test.toarray()})
            print(y_test.reshape(1, -1))
            print(predicts)

但是它总是返回准确的0。也许在定义变量和准确性时有一些错误。如何解决该问题并获得真正的准确性？

我检查测试数据并预测测试数据，它们看起来像

test -  [[0 0 1 ... 1 1 1]]
[[0.74700934]

test_predict - [0.59065396]
 [0.7749347 ]
 ...
 [0.6533902 ]
 [0.72667533]
 [0.67722285]]

Tensorflow：如何获得逻辑回归的准确性

0 个答案: