Time Series Analysis

Understand time series analysis, its core features, and why it is useful.

Defining time series analysis

The term time series refers to data in which observations are associated with a given moment in time. The evolution of Apple stock prices over the course of a day with the price per hour being the data point is an example of a time series. The daily temperature in Rio de Janeiro during 2021 is another example.

Analyzing time series data means trying to understand how our variable of interest behaves over time by identifying trends, seasonal patterns, and the influence of external factors such as holidays and weather.

How time series analysis is used

Time series analysis can be used to understand how something changes over time, but it's mostly used to make predictions. By identifying patterns, we can extrapolate them into the future and try to make an educated guess about how that variable will behave.

If we know the average temperature in Rio de Janeiro has been above 30 °C every year during January and February, we can infer that this will also be the case this year. As with any form of inference, this method is prone to error, though we will find out how to measure the amount of error we can expect according to the data we have available.

Here's an example of time series data up to 1961 with forecasts for the next year:

Example of time series forecasting
Example of time series forecasting

Content covered in the course

We'll learn everything from how to find time series data online to the basics of Python. We'll find out how to compose time series data and the main techniques used to make forecasts and assess them. By the end of this course, you’ll be able to start from scratch and build a forecasting model for any variable that you want, from stock prices to sales data.