Time Series Analysis of Alcoholic Beverages

Time Series Analysis of Alcoholic Beverages
Insights/Ouputs Rating
Average: 4.7 (3 votes)
Insight/Output Created On
Insight/Output Type
Insight/Output Description


In order to establish a better understanding on the amount of liquor people buy during previous decades, we obtain monthly data from FRED website from year 1992 to 2018, and construct the following time series project to analyse it. We utilize transformation and differentiation to gain the stationary data, and fit a SARIMA model possessing a low AICc and BIC value. Based on the diagnostic criterias and spectral analysis, we select the final best model and perform a forecast from Jan. 1st, 2017 to Dec.1st, 2017. Since the comparison between forecast data and observed data lied within the 95% confidence interval, it is reasonable to believe that our model is appropriate and accurate.