Learn time series forecasting
Eight short lessons that take you from "what is a time series?" to a forecast you can trust. Every lesson uses the same small example, so you can copy the code and run it as it is.
The lessons, in order
You only need basic Python and a little pandas. If you are new to pandas, start with the free pandas cheat sheet first.
- 1What is a time series?Understand what a time series is and spot its three building blocks.10 min
- 2Get your data ready in pandasTurn a messy file into a clean series with a proper date index.12 min
- 3Trend, season and stationaritySplit a series into its parts and make it stationary.15 min
- 4Your first forecast: baselines and smoothingMake three forecasts and compare them fairly.15 min
- 5ARIMA and SARIMAX, explained simplyFit a seasonal ARIMA model and read its forecast range.18 min
- 6Test your forecast honestlyTest any model on several time windows and pick the right error metric.15 min
- 7Machine learning and ProphetForecast with a machine learning model and with Prophet.18 min
- 8Practice projectsBuild a real forecasting project and show it with a clear story.20 min
Questions
Is the course really free?
Yes. Every lesson is free to read, with no sign-up and no email needed.
What do I need to follow along?
Python 3 with pandas, numpy, statsmodels and scikit-learn. Lesson 7 also uses Prophet, which is optional. Install them with pip.
Do I need to know statistics?
No. Each idea is explained in plain words first, then shown in code. The maths stays light.
Where is the cheat sheet?
The one-page time series forecasting cheat sheet is a free PDF that sums up all eight lessons.
How long does it take?
About two hours in total. Most lessons take 10 to 15 minutes, plus the time you spend trying the code.
Want a forecast built for your business?
I build forecasting tools and dashboards for small teams, and teach this material to university students.
Sales forecasting service