tf.contrib.timeseries

 

upper level

 

Modules

 

saved_model_utils:

Convenience functions for working with time series saved_models.

 

Classes

 

ARModel:

Auto-regressive model, both linear and non-linear.

ARRegressor:

An Estimator for an (optionally non-linear) autoregressive model.

CSVReader:

Reads from a collection of CSV-formatted files.

FilteringResults:

Keys returned from evaluation/filtering.

NumpyReader:

A time series parser for feeding Numpy arrays to a TimeSeriesInputFn.

RandomWindowInputFn:

Wraps a TimeSeriesReader to create random batches of windows.

StructuralEnsembleRegressor:

An Estimator for structural time series models.

TrainEvalFeatures:

Feature names used during training and evaluation.

WholeDatasetInputFn:

Supports passing a full time series to a model for evaluation/inference.

 

Functions

 

predict_continuation_input_fn(…):

An Estimator input_fn for running predict() after evaluate().