Summary

In this chapter, we introduced TSA as the process of extracting summary and other statistical information from time-series. We broke this process down into understanding the variables, uncovering relationships between variables, and identifying trend and seasonality.

We introduced datetime and pandas, the libraries sine qua non in TSA, and their functionalities for time-series; for example, resampling. Throughout the chapter, we listed and defined many summary statistics including mean, standard deviation, median, SE, confidence interval, Pearson correlation, and covariance.

We also talked about the concepts of seasonality, cyclic variation, and stationarity. We discussed why stationarity is important, and how to test for stationarity.

We also showed plotting functionality with Matplotlib and Seaborn, and how to generate different plots such as run charts, temporal line charts, correlation heatmaps, histograms, scatter plots, autocorrelation plots, and periodograms. In the practical example, we used an autocorrelation plot, which shows the correlation between different time steps, and the periodogram, which visualizes the power spectral density.