TensorFlow Estimator APIs Tutorials - TensorFlow v1.4

The tutorials use the TF estimator APIs to cover:

• Canned Estimators (Regression, Classification, Clustering, Time Series, Autoencoding, etc.)
• Metadata-driven approach to build the model features (work with numeric and categorical input attributes)
• Wide & deep Models - (handling dense and sparse feature columns)
• Scaling input feautures using the normalizer_fn in numeric_column()
• Feature Engineering (crossing, clipping, embedding, and bucketization, as well as custom logic during data input)
• Experiment APIs (tf.contrib.learn.experiment) and tf.estimator.train_and_evaluate (trainSpec & evalSpec)
• Data input (tf.estimator.inputs.pandas_input_fn, tf.train.string_input_producer, and tf.data.Dataset APIs)
• Work with .csv and .tfrecords (tf.example) data files
• Early Stopping (SessionRunHooks)
• Serving (export_savedmodel)
• Custom Estimators (model_fn & EstimatorSpec)