AttributeError: module 'tensorflow.contrib.learn' has no attribute 'TensorFlowDNNClassifier'

Question

This is the ml tensorflow code I am trying to execute -

import tensorflow.contrib.learn as skflow
from sklearn import datasets, metrics
iris = datasets.load_iris()
classifier = skflow.TensorFlowDNNClassifier(hidden_units=[10, 20, 10], n_classes=3)
classifier.fit(iris.data, iris.target)
score = metrics.accuracy_score(iris.target, classifier.predict(iris.data))

print("Accuracy: %f" % score)

It gives the following error -

Traceback (most recent call last):

File "C:\Users\admin\test3.py", line 5, in

classifier = skflow.TensorFlowDNNClassifier(hidden_units=[10, 20, 10], n_classes=3) AttributeError: module 'tensorflow.contrib.learn' has no attribute 'TensorFlowDNNClassifier'

[Finished in 69.3s with exit code 1]

[shell_cmd: python -u "C:\Users\admin\test3.py"]

martianwars martianwars · Accepted Answer · 2016-12-29T12:54:39

There seems to have been a major refactor in the TensorFlow project, and all the skflow code has been moved under the main tensorflow repository.

Try to replace TensorFlowDNNClassifier with just DNNClassifier. The new class can be found out here. Your corrected code will look like,

import tensorflow.contrib.learn as skflow
from sklearn import datasets, metrics
iris = datasets.load_iris()
# made a change in the next line
classifier = skflow.DNNClassifier(hidden_units=[10, 20, 10], n_classes=3)
classifier.fit(iris.data, iris.target)
score = metrics.accuracy_score(iris.target, classifier.predict(iris.data))

print("Accuracy: %f" % score)

AttributeError: module 'tensorflow.contrib.learn' has no attribute 'TensorFlowDNNClassifier'

2 Answers