4

I was trying to find the accuracy after training this simple linear model with sigmoid function: 

import numpy as np import tensorflow as tf import _pickle as cPickle with open("var_x.txt", "rb") as fp: # Unpickling var_x = cPickle.load(fp) with open("var_y.txt", "rb") as fp: # Unpickling var_y = cPickle.load(fp) with open("var_x_test.txt", "rb") as fp: # Unpickling var_x_test = cPickle.load(fp) with open("var_y_test.txt", "rb") as fp: # Unpickling var_y_test = cPickle.load(fp) def model_fn(features, labels, mode): # Build a linear model and predict values W = tf.get_variable("W", [4], dtype=tf.float64) b = tf.get_variable("b", [1], dtype=tf.float64) y = tf.sigmoid( tf.reduce_sum(W*features['x']) + b) if mode == tf.estimator.ModeKeys.PREDICT: return tf.estimator.EstimatorSpec(mode=mode, predictions=y) loss = tf.reduce_sum(tf.square(y - labels)) global_step = tf.train.get_global_step() optimizer = tf.train.GradientDescentOptimizer(0.01) train = tf.group(optimizer.minimize(loss), tf.assign_add(global_step, 1)) return tf.estimator.EstimatorSpec( mode=mode, predictions=y, loss=loss, train_op=train) estimator = tf.estimator.Estimator(model_fn=model_fn) x_train = np.array(var_x) y_train = np.array(var_y) x_test = np.array(var_x_test) y_test = np.array(var_y_test) input_fn = tf.estimator.inputs.numpy_input_fn( {"x": x_train}, y_train, batch_size=4, num_epochs=60, shuffle=True) estimator.train(input_fn=input_fn, steps=1000) test_input_fn= tf.estimator.inputs.numpy_input_fn( x ={"x":np.array(x_test)}, y=np.array(y_test), num_epochs=1, shuffle=False ) accuracy_score = estimator.evaluate(input_fn=test_input_fn["accuracy"]) print(accuracy_score)

But the dictionary doesn't have an "accuracy" key. How do I find it? Also, how do I use tensorboard to track the accuracy after each step?

Thank you in advance, the tensorflow tutorial is very bad at explaining.

Improve this question

3 Answers 3

Reset to default
2 
test_results = {} test_results['model'] = model.evaluate( test_features, test_labels, verbose=0) print(f" Accuracy: {test_results}")
Improve this answer
Sign up to request clarification or add additional context in comments.

1 Comment

Code-only answers are not particularly helpful. Please include a brief description of how this code solves the problem.
1

You need to create the accuracy yourself in model_fn using tf.metrics.accuracy and pass it to eval_metric_ops that will be returned by the function.

def model_fn(features, labels, mode): # define model... y = tf.nn.sigmoid(...) predictions = tf.cast(y > 0.5, tf.int64) eval_metric_ops = {'accuracy': tf.metrics.accuracy(labels, predictions)} #... return tf.estimator.EstimatorSpec(mode=mode, train_op=train_op, loss=loss, eval_metric_ops=eval_metric_ops)

Then the output of estimator.evaluate() will contain an accuracy key that will hold the accuracy computed on the validation set.

metrics = estimator.evaluate(test_input_fn) print(metrics['accuracy'])
Improve this answer

4 Comments

I did what you said but it throws the error "NameError: name 'predictions' is not defined".
You have to define it yourself like this: tf.argmax(y, axis=1)
In your case you should use predictions = tf.cast(y > 0.5, tf.int64) sorry (predict 1 when the output of sigmoid is above 0.5). I've updated the answer.
Link is broken, is the correct one this one? tensorflow.org/api_docs/python/tf/keras/metrics/Accuracy
0 
accuracy_score = estimator.evaluate(input_fn=test_input_fn) print(accuracy_score["loss"])

You can get loss like the above way for accuracy.

Improve this answer

1 Comment

doesn't work, the dictionary doesn't have the key "accuracy".

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.