Generating a synthetic time series dataset

In the previous recipes of this chapter, we trained and deployed models that deal with text classification and image classification requirements. In this recipe, we will generate a synthetic time series dataset similar to what is shown in Figure 8.23. This dataset will then be used later for training the DeepAR model in the recipe Training and deploying a DeepAR model.

Figure 8.23 – Time series plot

We can see that seasonal variations or seasonality are present in this time series dataset. At the same time, we can see that there is a bit of noise added to make the dataset a bit more realistic and enhance the robustness of trained machine learning models.

Getting ready

A SageMaker Studio notebook running the Python 3 (Data Science) kernel is the only prerequisite for this recipe.

How to do it…

The steps in this recipe focus on generating and plotting the synthetic time series dataset:

...