In Search of Forecasting

Comparing Seasonal Lag on Liquor Store Sales

Finding Seasonal Correlations

Liquor store data seems like a good test set for building out forecasting models. It should be highly seasonal, with holiday spikes and dry spells when quitting drinking is popular.

I am exploring the workhorse of forecasting - SARIMA. Seasonal Auto-Regressive Integrated Moving Average.

We see some dual spikes in late June and December

References

• https://otexts.com/fpp3/index.html