Statistical Learning Project 1

Assignment

Data set: proj1data.csv The data contain data of 1000 patients who had a particular liver operation. We want to use statistical learning method to find a statistical model for predicting survival in patients undergoing such operation.

• Response: Days: Survival time (in days).
• Predictor variables
  • Score: blood clotting score;
  • Index: prognostic index;
  • Enzyme: enzyme function test score;
  • Liver: liver function test score;
  • Age: age, in years;
  • Gender: gender(male, female);
  • Alcohol: alcohol use (0=None, 1=Moderate, 2=Severe).
  • Note: Gender and Alcohol are qualitative.
• Training Data: The first 500 cases of 1000 patients.
• Testing Data: The other 500 cases.
• First use the Box-Cox transformation MASS::boxcox() to find an appropriate transformation of the response variable. Then use the transformed variable as response \(Y\).
• Then try as many methods as possible that you have learned so far (at least those covered in Chapters 1 through 3, and 5, 6) and that are appropriate to this data set. The goal is to train at least three regression models including KNN regression and find a model that can be used to predict survival time using all or some of the predictor variables. The best final model should be chosen by comparing the test data means squared errors.

Project Report Format:

• Please typeset your report using R Markdown in RStudio to produce a PDF file. Your report should have the following

  • A title
  • Abstract
  • Introduction
  • Section(s) containing details of your data analysis, include only relevant R codes used for data analysis, graphs and explanation of the methods you used, etc. If random data were generated you must specified a random seed so that your results can be repeated.
  • Summary of Results and Discussion which include a table to summarize the test MSE’s of the models trained and give the predicting formula of your final model if possible.

• Limit your report within 15 pages. Only report necessary and relevant R code, graphs, and printout. Make your report neat, clean, readable, not like draft. Do not include the data in your report.

• Important: Using AI tools or anything like this is prohibited. Use only the R libraries and functions taught or mentioned in the lectures or the textbook or no credit!

• Submit both your PDF file and your source .RMD file on Canvas