SAS Homework 3

Homework 3

The dataset for this exercise is available in VideoGamesSales_Main.7bdat. This dataset contains information on the global sales and critic and user review ratings for videogames launched between 2001 and 2012 (from www.vgchartz.com). The variables are:

• Name of the game • Videogame platform on which it was released.

Platform DS Nintendo DS GBA Nintendo Game Boy Advance GC Nintendo Game Cube PC Personal Computer PS2 Sony PlayStation 2 PS3 Sony PlayStation 2 PSP Sony PlayStation Portable Wii Nintendo Wii X360 Microsoft XBOX 360 XB Microsoft XBOX

• Videogame Genre (e.g., Action, Sports, Shooter etc.) • Publisher • Developer • Rating: E = Everyone, E10+ = Everyone 10+, T = Teen, M = Mature • Global Sales (Millions of units) • Year of release • Critic Score (0 – 100): Average critic rating • Critic Count : Number of critic ratings • User Score (0 – 10): Average user rating • User Count: Number of user ratings

1. Develop a regression model (using proc reg) that links global sales to video game reviews. Explore ways in which the model fit could be improved through suitable changes to the model specification and variables. a. First, use proc freq to create a frequency table of 3 variables: platform, genre, and rating. b. In data step, create categorial variables for platform, genre, and rating using if/else statements.

Also create a variable for the age of the game relative to year 2013 (Note that these games were released before 2013).

c. Run a regression with all relevant X variables. Report the adjusted R-squared. d. Now, generate natural log of the following variables: global sales, critic_score, critic_count,

user_socre, user_count.

e. Run a regression with the log of Y variable and report adjusted R-squared. f. Run a regression with the log of Y variable as well as log of X variables generated in part d). Report

adjusted R-squared. g. Which model (out of part c, e, and f) offers the highest adjusted R-squared? What would be the

economic reasoning on why that particular model provides the best fit? Also interpret coefficients.

2. Replicate your final model in Q.1 part g) using proc glm and class commands. Note that proc glm allows

you quickly check many basic results. However, it does not provide many diagnostic outputs that proc reg provides.

a. Interpret the parameter estimates for each of ‘genre’ in plain English. b. Interpret the parameter estimate for ‘rating’ in plain English. c. Interpret the parameter estimate for ‘lg_user_count’ in plain English.