Our data is properly formatted and we can now train our model. For this task, we are using LSTM. This is a particular type of RNN. These types of neural networks are a good choice for time-series data because they are able to take time into account during the modeling process.
Most neural networks are classified as feedforward networks. In these model architectures, the signals start at the input node and are passed forward to any number of hidden layers until they reach an output node. There is some variation in feedforward networks. A multilayer perceptron model is composed of all dense, fully connected layers while a convolutional neural network includes layers that operate on particular parts of the input data before arriving at a dense layer and subsequent output layer. In these types of models, the backpropagation step passes back derivatives...
