Zouwu AutoTS Quickstart

In this guide, we will show you how to use AutoTS for automated time series forecasting.

The general workflow using AutoTS contains below two steps.

  1. create a AutoTSTrainer to train a TSPipeline, save it to file to use later or elsewhere if you wish.
  2. use TSPipeline to do prediction, evaluation, and incremental fitting as well.

Refer to AutoTS notebook for demonstration how to use AutoTS to build a time series forcasting pipeline.

Refer to AutoTS API Guide for more details of AutoTS APIs.

Step 0: Prepare environment

We recommend you to use Anaconda to prepare the enviroments, especially if you want to run automated training on a yarn cluster (yarn-client mode only).

conda create -n zoo python=3.7 #zoo is conda enviroment name, you can set another name you like.
conda activate zoo
pip install analytics-zoo[automl]==0.9.0.dev0 # or above

Step 1: Init Orca Context

You'll need RayOnSpark for training with AutoTSTrainer, so you should init an orca context with init_ray_on_spark=True before auto training, and stop it after training is completed. Note orca context is not needed if you just use TSPipeline for inference, evaluation or incremental training.

from zoo.orca import init_orca_context, stop_orca_context

# run in local mode
init_orca_context(cluster_mode="local", cores=4, memory='2g', num_nodes=1, init_ray_on_spark=True)

# run in yarn client mode
                  num_nodes=2, cores=2, 
                  driver_memory="6g", driver_cores=4, 

Step 2: Create an AutoTSTrainer

To create an AutoTSTrainer. Specify below arguments in constructor. See below example.

 from zoo.zouwu.autots.forecast import AutoTSTrainer

 trainer = AutoTSTrainer(dt_col="datetime",

Step 3: Fit with AutoTSTrainer

Use on train data and validation data. A TSPipeline will be returned.

 ts_pipeline =, validation_df)

Both AutoTSTrainer and TSPipeline accepts data frames as input. An exmaple data frame looks like below.

datetime value extra_feature_1 extra_feature_2
2019-06-06 1.2 1 2
2019-06-07 2.3 0 2

Note: you should call stop_orca_context() when your distributed automated training finishes.

Step 4: Further deployment with TSPipeline

Use to train pipeline (incremental fitting), evaluate or predict.

 #incremental fitting, new_val_df, epochs=10)

Use to load from file or save to file.

 from zoo.zouwu.autots.forecast import TSPipeline
 loaded_ppl = TSPipeline.load(file)
 # ... do sth. e.g. incremental fitting