This paper serves to address the effect of time on the sales of clothing retail, from 2010 to May 2019. The data was retrieved from the US Census, where N=113 observations were used, which were plotted to observe their trends. Once outliers and transformations were performed, the best model was fit, and diagnostic review occurred. Inspections for seasonality and forecasting was also conducted. The final model came out to be an ARIMA (2,0,1). Slight seasonality was present, but not enough to drastically influence the trends. Our results serve to highlight the economic growth of clothing retail sales for the past 8 years, cementing the significance of the production economy's stability. The quarterly GDP data was collected in order to find out the relationship with the differenced clothing data. Some observations of GDP data were affected by the clothing data before removing the seasonality. After removing the seasonality, the clothing expense is white noise and not predictable from the historical GDP.
Bachelor of Science (B.S.)
College of Sciences
Statistics and Data Science
Huang, Weijun, "Time Series Forecasting and Analysis: A Study of American Clothing Retail Sales Data" (2019). Honors Undergraduate Theses. 643.