Forecasting with Holt's Linear Trend

Simple exponential smoothing can't follow a trend: forecasts lag behind growing sales. Holt's method tracks both a level $\ell$ and a trend $b$ :

\ell_t = \alpha y_t + (1-\alpha)(\ell_{t-1} + b_{t-1}), \qquad b_t = \beta(\ell_t - \ell_{t-1}) + (1-\beta)\,b_{t-1}

Initialise with $\ell_0 = y_0$ and $b_0 = y_1 - y_0$ , then update for $t = 1 \dots n-1$ . The forecast $h$ steps after the last observation is $\ell_{n-1} + h \cdot b_{n-1}$ .

Implement holt_forecast(y, alpha, beta, horizon) returning a list of horizon forecasts. Raise ValueError if y has fewer than 2 points. Then implement mape(actual, forecast), the mean absolute percentage error as a percentage, skipping points where the actual value is 0.