What you'll build

You'll forecast Cyclepath, a city bike-share's monthly ridership, across the whole course — decomposing it into trend and seasonality, testing it for stationarity, fitting ARIMA and SARIMA models, comparing exponential-smoothing baselines, and backtesting your forecasts before making a final prediction. The data is seeded synthetic with a clear trend and yearly seasonality, generated in-course, and every model is fit for real with pandas, numpy, and statsmodels — fully reproducible, no API key, no cost. This is the classical, statistical approach to forecasting (ARIMA and friends), the complement to the deep-learning sequence models in our ML program.

Before you start

You'll need comfortable Python with pandas and numpy, and a first grounding in statistics — means, variance, and the idea of a hypothesis test. Our Statistics & Probability and Machine Learning Foundations courses are ideal preparation. This course teaches the classical/statistical forecasting methods (ARIMA, SARIMA, exponential smoothing); the deep-learning sequence approach (RNNs/LSTMs) is covered separately in our deep learning track.

Set up your environment

You can complete this course on any machine with Python 3.10+. There's no API key and nothing to pay for — every series is generated locally with a fixed random seed.

1. Install the packages the course uses:

pip install pandas numpy statsmodels matplotlib

Every dataset is generated in the lesson with a fixed seed, so your series and model outputs will match the ones shown, and you can rerun any forecast end to end.

Package APIs shift over time. If a statsmodels signature has changed since these versions, the forecasting concepts still apply — adjust the syntax to what you have installed.