Using that$$stream$$of data$$and the same assumptions about handling the smoothed and trending portions of the early periods of the data stream,$$optimize the forecast by sequentially: $1)$finding the alpha$($to the nearest increment of$0\mathrm{.}05)$that produces the lowest Se$,$ then$2)$holding that alpha constant,$$and finding the beta$($to the nearest$0\mathrm{.}05)$which provides the overall lowest Se$.$Remember$,$when changing alpha or beta using this model,$$you must change the entire column, not$$just one row at a time.

$$ What were the final optimal alphas and betas?

$$What is the eventual lowest forecast error using this process?

$$How might you proceed to find a forecast with even lower error?

$$What is the point estimate for the next unknown period?$$What is the$95\%$confidence$$interval estimate?