Apache Spark Deep Learning Cookbook

How it works...

This section explains how the LSTM neural network model is configured to train on our dataset.

1. Most of the functionality from keras used to build the LSTM model will come from models and layers.
2. The LSTM model that has been built will be defined using a Sequential class that works well with time series that are sequence dependent. The LSTM model has an input_shape = (1,5) for one dependent variable and five independent variables in our training dataset. Only one Dense layer will be used to define the neural network as we are looking to keep the model simple. A loss function is required when compiling a model in keras, and since we are performing it on a recurrent neural network, a mean_squared_error calculation is best to ...