Summary

In this chapter, we attempted to analyze and understand time series data, as well as developing two predictive forecasting models in less than 15 pages. We began our journey by exploring and decomposing time series data into smaller features that can generally be used with shallow machine learning models. We then investigated the components of a time series dataset to understand the underlying makeup. Finally, we developed two of the most common forecasting models that are used in the industry – the Prophet model, by Facebook, and an LSTM model using Keras.

Throughout the last few chapters, we have developed various technical solutions to solve common business problems. In the next chapter, we will explore the first step in making models such as these available to end users using the Flask framework.