• The data shows retail sales at hardware stores (in millions of dollars) in the United States monthly and are not seasonally adjusted. a) Use ForecastX and display the first 24 autocorrelations for the series. What do the ACF and PACF suggest about the series? Explain the patterns you see (commentary required) b) Does differencing make sense to use with the series. Explain why or why not (commentary required) c) Estimate an ARIMA model that you believe to be a good candidate for forecasting future retail sales at hardware stores. Test the Ljung-Box statistic and report your findings. Finally, plot the first 24 autocorrelations of the residuals to your best model. d) What conclusions do you draw about your model using Ljung-Box test, autocorrelation plot of residuals and model statistics (commentary required)