Forecasting

Forecasting means prediction!!!

“Predicting a single number tells you nothing because we are dealing with probabilities and randomness, so always produce a confidence interval, even in conversation” — Reza the Goat

1. Pick a model,
2. then residual diagnostics to evaluate fit and model assumptions, i.e. $N(0,\sigma )$ normality or residuals
3. if residual diagnostics pass you THEN do inference (prediction, confidence intervals, p-values).

\item $Cov(X_i,X_j)=0$ for uncorrelated RVs

\item for two RVs, Independence is a stronger condition than non-correlation

\item WHAT ARE DEGREES OF FREEDOM?