Too much Human Growth Hormone released from the Pituitary fo…
Questions
Tоо much Humаn Grоwth Hormone releаsed from the Pituitаry for too long results in...
Use the “Prоblem1” dаtаset tо cоmplete this question. The dаta set contains monthly sales (in dollars) for a souvenir shop between 1995-2001. Partition the data into training and validation sets, with the validation set containing the last 12 months of data (year 2001). 1.1 Plot the training time series. (8 pts) 1.2 Run a regression model with Sales as the outcome variable, and with a linear trend and monthly seasonality. Call this modelA. Show the model summary. (5 pts) 1.3 Run another regression model with log(Sales) as the outcome variable, and with a linear trend and monthly seasonality. Call this modelB. Show the model summary. Hint: fitting a linear model to log(Sales) is equivalent to fitting an “exponential” trend to Sales. (5 pts) 1.4 With modelB in (1.3), interpret the effect of the trend. Interpret the coefficient for “Season12” (December). (6 pts) 1.5 Compare the fit between modelA in part (1.2) and modelB in part (1.3) on the training data. Mention at least two reasons. (6 pts) 1.6 Use modelB in part (1.3) to forecast monthly sales in year 2001. Show point forecasts for year 2001 from January to December. (5 pts) 1.7 Plot the entire time series, overlayed with the predicted series in the training period and forecasted series in the validation period. Use different colors and line types for the three series. Add legend for the three sets of time series in the plot. (7 pts) 1.8 Create an ACF (autocorrelation) plot for the regression residuals of modelB in (1.3) with showing 12-lag periods. What is the value of lag-1 autocorrelation? (5 pts) 1.9 Fit an AR(1) model on the residuals of modelB in (1.3). Show model summary. Write down the estimated AR(1) model. (5 pts) 1.10 Use the model in (1.9) to adjust the forecast for January 2001 given by modelB in part (1.3). Give the value of the adjusted forecast. What is the actual value of sales in January 2001? Note that the residuals from modelB is on a log scale. (8 pts)