Time Series Revenue Forecasting AI Prompt

Build a statistical time series forecast for this revenue or financial metric.

Metric: {{metric}}
Historical data: {{data}} (at least 24 periods)
Forecast horizon: {{horizon}} periods
Seasonality: {{seasonality}} (yes/no, and period: weekly/monthly/quarterly)

1. Exploratory analysis:
   - Plot the series: trend, seasonality, and irregular components visible?
   - Decompose using STL or classical decomposition: trend + seasonal + residual
   - Autocorrelation function (ACF) and Partial ACF (PACF): identify autoregressive and moving average structure
   - ADF test: is the series stationary? If not, difference and retest.

2. Model candidates:

   Holt-Winters (ETS):
   - Triple exponential smoothing: handles trend and seasonality
   - Parameters: alpha (level), beta (trend), gamma (seasonality)
   - Best for: smooth trends with regular seasonality, limited data (< 5 years)

   SARIMA:
   - Seasonal ARIMA (p,d,q)(P,D,Q)[m]
   - Select parameters using auto.arima (AIC/BIC minimization)
   - Best for: data with complex autocorrelation structure

   Prophet (Facebook/Meta):
   - Handles multiple seasonalities, holiday effects, and trend changepoints
   - Best for: monthly or daily data with irregular events and holidays

3. Model evaluation:
   - Split data: train on first 80%, test on last 20%
   - Metrics: MAPE, MAE, RMSE on the test set
   - Select the model with lowest MAPE on holdout

4. Forecast output:
   - Point forecast for each future period
   - 80% and 95% prediction intervals
   - Is the interval width reasonable given the metric's historical variance?

5. Forecast decomposition:
   - How much of the forecast is trend vs seasonal vs baseline?
   - What is the year-over-year growth implied by the forecast?

Return: model comparison table, selected model diagnostics, forecast with prediction intervals, and decomposition of forecast components.