To get a sense of the data, provide two well labeled scatter-plots of 1) revenue vs. budget, and 2) revenue vs. screens. Also, provide a correlation matrix of these three variables
b) After speaking to an industry expert, they tell you that budget and screens are the most important determinants of film success at the box office (advertising is important too, the expert tells you, but you don’t have data on that). Produce a regression with ‘revenue’ as the dependent variable and ‘budget’ and ‘screens’ as the independent variables. Interpret the regression coefficients.
c) With your regression results of part b), provide residual plots vs. each independent variable. Also, provide a histogram of the (standardised) residuals. Comment on the nature of the residuals based upon these three plots.
d) You show your results and plots to your econometrics professor at university and he tells you that you may have an issue with non-normal errors caused by outliers. He suggests you transform the revenue, screens, and budget data into natural logarithms and re-estimate the equation, which you do. Provide results of the new regression and interpret the regression coefficients [Hint: d lnY/d lnX = (dY/dX)*(X/Y) = (dY/Y)/(dX/X)].
e) With your new regression results of part d), provide residual plots vs. each independent variable. Also, provide a histogram of the (standardised) residuals. Comment on the nature of the new residuals based upon these three plots.
Document Preview: Example Regression Output