Final Project 8

Project Environment Setting Up

1. Clone the Repository from the github to your local machine
2. Navigate to the project directory in terminal

Report Generation

There are two ways to generate the report: locally on your computer or using a Docker image.

• Locally:

1. Make sure you have make and R installed on your system
2. Make sure you have renv R package is installed
3. Open a terminal in the project directory
4. Run command make install to restore the R package environment using renv
5. Run make report to compile the final report

• Using Docker:

1. Pull the image from DockerHub Repository
2. Run command make mount-report in the terminal to generate the report
   (This step works for both Windows-OS and a Mac/Linux-OS)
3. The compiled report should be in your local \report folder
4. (Optional) If you prefer to build the image yourself instead of downloading it from DockerHub, use the command make build_image. An image called "wwwivy111/data550_final_project" will be built.

Repository stucture

The raw dataset Thyroid_Diff.csv was saved in the data/ folder.
Codes were saved in the code/ folder.
The final report was saved in the report/ folder.

• README.md
• Makefile
• Dockerfile
• renv.lock
• renv/
• data/
• code/
• output/
• report/

Report contents

Key sections include:

1. Introduction: Introduce the study
2. Method and Analysis: Describes the methods and results for data preparation, exploratory data analysis, and modeling process and model evaluation
3. Discussion: Discusses the implications for clinical management and future research

Code description

code/01_split_data.R

• cleans the data format
• splits the data into train and test set
• saves new datasets as different .rds objects in data/ folder
  (clean_data.rds, train.rds, test.rds)

code/02_EDA.R

• conducts Exploratory Data Analysis (EDA)
• generates table1 and saves as table1.rds object in output/ folder
• generates descriptive plots for outcome, continuous, and categorical variables and saves as .png objects in output/ folder
  (descriptive_age_plots.png, descriptive_bar_outcome.png, descriptive_pie_charts.png)

code/03_modeling.R

• generate new train and test data train_1.rds and test_1.rds in data/ folder
• fits univariate models and multivariable model, stepwise selection model, and final model
• saves models and corresponding tables as different .rds objects in output/ folder
• conducts model evaluation for the final model
• saves evaluation matrix and ROC plot as .rds and .png objects in output/ folder

code/04_render_report.R

• renders report.Rmd

report.Rmd

• reads outputs from code/01_split_data.R, code/02_EDA.R, code/03_modeling.R
• makes the final report

Makefile

• contains rules for building the final report and other targets
• make report will compile the report into .html object
• make split_data will generate the outputs of code/01_split_data.R
• make EDA will generate the outputs of code/02_EDA.R
• make modeling will generate the outputs of code/03_modeling.R
• make clean will clean all outputs