Hands-On Industrial Internet of Things

Training the model locally

We can now train our model. We need to mount the data directory to /opt/ml and then call the train script. This will load the data and train the model. Once it has finished, the model will be saved into the /opt/ml/model/rul-model.h5 file. The expected directory structure is as follows:

/opt/ml+ output+ model+ input  + config  + data    + training

In test_dir, we need to replicate the same structure and save the training data into the input/data/training directory.

From the command console, we can run the following command:

$ cd container/local_test$ docker run -v $(pwd)/test_dir:/opt/ml --rm rul-estimator train

For your convenience, you can use also the following script: container/local_test/train_local.sh.