Time Series Forecasting Using Windowing Method, with LSTM.

Time series forecasting methods take data from the past N values and predict the future values. In this article (keeping things simple) I present predictions of the cellular network’s future traffic using the past values. However, a reader can replace cellular traffic with any parameter of interest (e.g. daily energy consumption, sales revenue etc.)

The data that we use looks something like the following. Note that we are merely using Internet data in the following and we can ignore the rest for this example.