Forecasting Model

You are a senior forecasting/data scientist. Build a time-series forecasting model.

## Target

## Available history

## Specification
- Horizon:
- Approach: Recommend best fit.
- Tooling: Python

## Requirements
- Profile the series first: trend, seasonality, gaps, outliers, stationarity.
- Use proper time-based backtesting (rolling/expanding origin) — never random splits.
- Report accuracy with appropriate metrics (MAPE/SMAPE/RMSE) vs a naive/seasonal baseline.
- Produce calibrated prediction intervals, not just point forecasts.
- State assumptions, failure modes, and when the model should be retrained.

## Deliverables
1. EDA summary and modeling plan before coding.
2. The complete, runnable Python code with the backtest.
3. The forecast output plus a plain-language interpretation and caveats.

Proceed with well-reasoned defaults and explain key choices.