Extracting features automatically from a time series
Time series are data points indexed in time order. Based on a time series sequence, we can predict several things. For example, we can predict whether a pipeline will fail, a music genre, whether a person is sick, or, as we will do in this recipe, whether an office is occupied.
To train regression or classification models with traditional machine learning algorithms, we require a dataset of size M x N, where M is the number of rows and N is the number of features or columns. However, with time series data, what we have is a collection of M time series. And each time series has multiple rows indexed in time. To use time series in supervised learning models, each time series needs to be mapped into a well-defined feature vector, N, as shown in the following diagram:
Figure 10.1 – Diagram showing the process of feature creation from a time series for classification or regression