Zaman Serileri ve Tahmin

Classical Forecasting

1. Time Series Structure: Trend, Seasonality, StationarityDecomposing a series into trend, seasonal, and remainder parts, and checking stationarity with autocorrelation.1–2 sa
2. Autoregressive and ARIMA ModelsForecasting from lagged values and lagged errors, with differencing for non-stationary series.1–2 sa
3. Exponential SmoothingWeighted averages that decay with age, extended to trend and seasonality (Holt-Winters).1–2 sa

Forecasting Practice

1. Probabilistic ForecastingForecasting distributions rather than points: prediction intervals, quantiles, and proper scoring rules.1–2 sa
2. Forecast Evaluation and BacktestingTime-respecting splits, rolling-origin evaluation, and scale-free error metrics.1–2 sa
3. Deep Learning for ForecastingGlobal neural forecasters across many series, and the evidence on when they beat classical baselines.1–2 sa