Tutorial – forecasting demand using Prophet and LSTM
In this tutorial, we will use the sales dataset from the previous section to develop two robust demand forecasting models. Our main objective will be to use the sales data to predict demand at a future date. Demand forecasting is generally done to predict the number of units to be sold on either a given date or location. Companies around the world, especially those that handle temperature-sensitive or time-sensitive medications, rely on models such as these to optimize their supply chains and ensure patient needs are met.
First, we will explore Facebook’s famous Prophet library, followed by developing a custom Long Short-term Memory (LSTM) deep learning model. With this in mind, let’s go ahead and investigate how to use the Prophet model.
Using Prophet for time series modeling
Prophet is a model that gained a great deal of traction within the data science community when it was first released in 2017...