Strengths: A range of techniques were used carefully; a nicely written report.

Points for consideration:

  1. AIC cannot readily compare likelihood of data with different amounts of differencing.

  2. Seasonality is likely in sales and production. If seasonality is the main question, it would be nice to see which holidays match measurable seasonal peaks (e g., Halloween and/or Christmas and/or Easter and/or 4th July).

  3. What does auto.arima() do? We learn more from comparing AIC values than simply which is lowest.