Time Series
Time series breaks the i.i.d. assumptions of ordinary ML: order matters, observations correlate with their own past, and you can't shuffle. Learn to decompose, test for stationarity, and forecast with the classical workhorses (AR, MA, ARIMA, SARIMA, SARIMAX, VAR) and modern tools (Prophet) — with honest, leakage-free evaluation.
- Chapter 01
Foundations
4 lessons - 01 Why time series is different Why you cannot treat time-stamped data like an ordinary tabular dataset — and what that forces you to do differently.
- 02 Trend, seasonality & decomposition Learn to separate the long-term trend, repeating seasonal pattern, and random noise in any time series — and see why that split is the first step in almost every forecasting workflow.
- 03 Stationarity, ADF & differencing ARIMA keeps demanding a 'stationary' series. What does that mean, how do I check it, and how do I make my data stationary?
- 04 ACF & PACF How ACF and PACF plots reveal the AR and MA order fingerprints that guide ARIMA model selection.
- Chapter 02
Classical Models
6 lessons - 05 Autoregression (AR) Learn how the AR(p) model explains tomorrow as a weighted blend of recent history, and why the PACF tells you exactly how many lags to include.
- 06 Moving average (MA) How MA(q) models the lingering effect of past shocks — and why it is nothing like a rolling-average smoother.
- 07 ARIMA Learn how autoregression, differencing, and moving-average error terms combine into one unified model you can fit, diagnose, and forecast with.
- 08 SARIMA (seasonal) Extend ARIMA with a second set of seasonal AR, differencing, and MA terms that operate at lag m — the season length — so the model captures yearly, weekly, or any repeating cycle.
- 09 SARIMAX (exogenous regressors) Add holiday flags, promo spend, price, and other known external drivers to a seasonal ARIMA model — and learn why forecasting with them requires knowing the future values of those drivers.
- 10 VAR (multivariate) Model several interdependent time series at once — interest rates, inflation, and unemployment — so that every variable's forecast draws on the past of every other variable in the system.
- Chapter 03
Modern & Practical
4 lessons - 11 Exponential smoothing & Holt-Winters Learn how Simple Exponential Smoothing, Holt's method, and Holt-Winters build up a fast, robust forecaster by weighting recent data more heavily — and when this ETS family beats ARIMA.
- 12 Prophet Learn how Meta's Prophet decomposes a time series into trend, seasonality, and holiday effects — and why its analyst-friendly API makes it a go-to for business forecasting.
- 13 Lag & rolling features How to turn a time series into a plain feature table so any regressor — XGBoost, LightGBM, linear models — can forecast the next value.
- 14 Evaluating forecasts (walk-forward) How to honestly measure whether your forecast is any good — and prove it beats just guessing yesterday's value.
- End of section 0 / 14 complete
Make it stick — pass every quiz.
Each lesson has a short quiz at the bottom. Passing the quiz is what marks the lesson complete and counts toward your certificate.
Section complete 14 / 14 lessonsNice work — you finished Time Series.
Certificates are earned per learning path, not per section. Here's where this section takes you:
Pick a learning path to start working toward a certificate.
Time Series — frequently asked questions
Straight answers to the questions people ask most about time series.
Why can't I use a normal train/test split for time series?
Because time series observations are ordered and correlated with their own past, so randomly shuffling them leaks future information into training and inflates your scores. Always split by time — train on earlier data and test on later data — and validate with walk-forward backtesting rather than random k-fold cross-validation.
Read the lessonWhat is the difference between ARIMA and SARIMA?
ARIMA models trend and short-term structure through autoregression, differencing, and moving-average terms but assumes no repeating seasonal pattern; SARIMA adds a seasonal set of those same terms at the season's period (for example every 12 months) to capture cycles ARIMA misses. Use SARIMA when your data has a clear, fixed-length seasonal pattern.
Read the lessonWhat does it mean for a time series to be stationary?
A stationary series has statistical properties — its mean, variance, and autocorrelation — that stay constant over time, which is exactly what models like ARIMA assume. You check it with a plot plus the Augmented Dickey-Fuller test, and usually achieve it by differencing the series to remove trend and seasonality.
Read the lessonWhen should I use Prophet instead of ARIMA?
Reach for Prophet when you want strong seasonality, holiday effects, and robustness to missing data and outliers with minimal tuning, since it is designed for business forecasting by analysts. Prefer ARIMA or SARIMA when you want a well-understood statistical model and are willing to identify orders from the ACF and PACF — and check either one against a naive baseline.
Read the lesson