Introduction
Extra Credit Data Analysis
Ends November 29
In this extra credit assignment, you will be tasked with building a predictive model. You will form teams of no more than four members. By signing up for this assignment, you are committing to putting forth a genuine effort to win the competition.
The training dataset contains 225 covariates for 300 observations, but many of these covariates are useless in predicting the response. You may also need to transform some of the variables to obtain a linear relationship. I also provided you with a testing dataset with 50 observations. I recommend using your training dataset to find an appropriate model (using, say, 10-fold cross-validation or PRESS) and then seeing how well your model works on the testing set.
I will retain a second testing dataset that will contain the same 225 covariates for another 50 observations. Your group will make predictions for these 50 observations. The winning group will have predictions that generate the lowest root mean-squared error for prediction:
1 50 root−MSEP= (yi−yˆi)2
50 i=1
Important Facts:
• The winning group will receive 5 extra points added to their next exam (5 extra points
on a 50 point exam = 10%)
• At the end of each week, your group is required to submit your current pre- dictions. I will calculate the root-MSEP for each group and post a leader board.
• At the end of the competition, your group will submit a short write-up that briefly sum- marizes your methods and final model.
• You are not to share any of your results with other groups.
• You may ask me questions about R packages and conceptual questions.
1