Rr-statistics.co

Courses ▾Tutorials Practice Pricing Tools

Get Certified →Sign in

Forecaster

Certified R Forecaster

marks a Pro lesson

Course overview

Curriculum

01Time series foundations8

What makes time series different Tidy temporal data with tsibble What a time series says before you model it EDA for time series Seasonal, subseries and lag plots Autocorrelation and the ACF Train and test splits for temporal data The benchmark forecasts every model must beat

02Decomposition and features6

Moving averages and classical decomposition STL decomposition in practice Seasonal adjustment with X-13 and STL Box-Cox and variance-stabilizing transforms Calendar and population adjustments Feature extraction with feasts

03The forecasting toolbox6

Fitted values, residuals and innovations Residual diagnostics and the Ljung-Box test Forecast distributions and prediction intervals Point forecasts versus the whole distribution Time-series cross-validation with rolling origins Accuracy metrics: MASE, RMSSE and pinball loss

04Time series regression6

Fitting a trend line with TSLM Trend and seasonal dummy variables Fourier terms for seasonality Lagged predictors, calendar effects and holiday dummies Selecting predictors with cross-validation Spurious regression and how to avoid it

05Exponential smoothing (ETS)6

Simple exponential smoothing Holt's linear trend and the damped trend Holt-Winters seasonal smoothing, additive and multiplicative The ETS taxonomy: what the three letters mean How to let ETS() choose a model, and when to override it How to forecast from a fitted ETS model, and check whether it beats the baselines

06ARIMA and seasonal ARIMA7

Testing whether a series is stationary with KPSS Removing trend and season with differencing Reading the ACF and PACF to tell AR from MA Fitting a non-seasonal ARIMA model and reading its coefficients Fitting and forecasting a seasonal ARIMA model Choosing between ARIMA and ETS Automatic and manual ARIMA selection

07Dynamic regression and complex seasonality6

Fitting a regression with ARIMA errors Fitting a dynamic harmonic regression Fitting Fourier terms for two seasonal periods Fitting a TBATS model for complex seasonality How Prophet fits trend, seasonality and holidays Lagged predictors, transfer functions, and forecasting the predictor first

08State space models and the Kalman filter7

An observed series as a noisy view of a hidden state The predict-then-update loop inside the Kalman filter Local level and local linear trend models Level, slope, seasonal and irregular in one fitted model A posterior forecast band instead of one fitted line Dynamic linear models: a coefficient that drifts over time Choosing between a rolling window and a state space filter

09Spectral and frequency-domain analysis5

The frequency domain and the periodogram Estimating the spectral density Low-pass, high-pass and band-pass filtering Detecting hidden periodicities Wavelets for non-stationary signals

10Volatility modeling: ARCH and GARCH6

Conditional heteroskedasticity and volatility clustering ARCH Models The GARCH family of volatility models Fitting GARCH in R Forecasting volatility and Value at Risk Multivariate volatility and DCC-GARCH

11Multivariate time series6

Fitting a vector autoregression (VAR) to two related series Choosing the Lag Order Testing Granger causality between two series Tracing a shock through a VAR with impulse responses Cointegration and the VECM Forecasting multiple related series

12Hierarchical and grouped forecasting5

Hierarchical and grouped time series Bottom-up, top-down and middle-out Optimal reconciliation with MinT Reconciliation with fabletools Coherent forecasts finance can plan against

13Machine learning and deep forecasting7

One global model across many series Machine Learning Forecasting with modeltime Boosted Trees for Forecasting Forecasting with a neural network using NNETAR How DeepAR, N-BEATS and N-HiTS forecast How a Temporal Fusion Transformer forecasts Bagging and combining forecasts

14Intervention, causal impact and anomaly detection5

Intervention analysis for a known event Estimating an event's effect with a counterfactual forecast Finding changepoints when the date is unknown Detecting anomalies and outliers in a series Handling outliers in a time series

15The expert edge: forecasting like a pro7

Rolling-origin backtests without leakage Conformal prediction intervals for time series Forecasting intermittent demand with Croston, ADIDA and TSB Forecast value added against a naive baseline Scaled errors and what the M competitions found Judgmental adjustments and forecast governance Checking prediction interval coverage

16Production forecasting6

Choosing how often to refit a forecast model Monitoring forecast accuracy over time Diagnosing a drop in forecast accuracy Calendar and event features for production forecasts Forecasting thousands of series with fable Deploying a forecasting service

Home/Tracks/Forecaster

What makes time series different

Sign in to run this lesson in the page.

Pro

Continue learning

Pro opens the rest of this track, every other track on the site, the graded practice and Certified R Forecaster. The first lesson of every section stays free whether you buy or not.

Unlock full access toNew to RData AnalystData ScientistForecasterResearcherR EngineersoonML Engineersoon

View all plans · Already Pro? Sign in

30% off

You have unlocked a limited-time coupon

Applies to the annual plan and Lifetime. Monthly renews at its displayed price.

CODE

Use it before it expires in

60:00

Use 30% coupon →

This offer is issued once for your account and will not appear again.

© 2016–2026 r-statistics.co

Tutorials Tools Pricing Verify a credential About