1 Data Set Description

This data comes from the United States Census Bureau and contains the monthly Sales from “hobby” products in the United States from 1992 to 2023.Some examples of products considered “hobby” include books, sporting goods, and musical instruments. The data is not seasonally adjusted and presented in a scale of millions of dollars.

2 Initial Time Series Graph

The initial time series object was built using the data from 1992 to 2021, the remainder of the data points were partitioned out for testing. The graph of the times series shows both a positive overall trend and strong seasonality, with December being a particular high point for hobby sales. There is a notable anomalous drop in sales during 2020 that can be attributed to the covid-19 pandemic.

3 Forcasts Graphs

The following four graphs show the forecasts for the next 20 months based on 4 baseline forecasting methods. The 4 forecasting methods are moving average, naive, seasonal naive, and drift respectively.

4 Testing Accuracy of Forcasting Methods

By calculating some standard accuracy measurements for the 4 forecasting methods, it becomes clear that the seasonal naive method is by far the most accurate. Seasonal naive method manages has by a wide margin the smallest mean absolute predictive error, mean absolute deviation, and mean square error. This is not a surprising result as the seasonal naive method is the one method the considers seasonality in its forecasting. Seasonality seems to have a heavy influence on hobby sales, much greater than any long term trends.

Overall performance of the four forecasting methods
MAPE MAD MSE
Moving Average 27.342412 48259.74 7154531
Naive 19.321137 34841.00 4329253
Seasonal Naive 3.796224 6049.00 145446
Drift 18.158241 32840.88 3951104