Create intermediate Works with: Any AI

Time Series Forecasting Baselines

Table output

Beat the naive forecast honestly.

The Open buttons send the prompt text as currently shown below to that site only when you click. Fill in the placeholders first.

Set up a forecasting evaluation. Series description and frequency: [SERIES_DESCRIPTION]. Horizon needed: [HORIZON]. Known events and seasonality: [EVENTS_AND_SEASONALITY]. Library: [LIBRARY].

Provide:
1. Baselines to beat: last value, same value last season, moving average, and seasonal naive. Give the code for each.
2. A proper backtest: rolling-origin evaluation with an expanding window, with a diagram of folds in text, and a warning against random cross-validation for time series.
3. Metrics: MAE, RMSE, and a scaled error such as MASE, with formulas, and why MAPE misbehaves near zero.
4. Candidate models: exponential smoothing, ARIMA family, gradient boosting with lag features, and a prophet-style additive model, with a one-line description of when each helps.
5. Prediction intervals and how to check their coverage (do 80 percent of actuals fall inside an 80 percent interval?).
6. A results table template and a decision rule for choosing the model.
7. Communication: how to present the forecast as a range with assumptions.
Mention what to do with holidays, outliers, and structural breaks.

Fill in before use

Optional: type your details and the prompt above updates. Nothing is saved or sent.

Replace each bracketed placeholder with your own details, then paste into Any AI or another assistant.

• forecast-baselines