praccomp2024

PCA of Hurricane Effects on Oceanographic Data

This project applies Principal Component Analysis (PCA) to oceanographic buoy data. When the resutling PCs are used with Generalized Linear Models (GLMs) and fish abundance data, they can be used to investigate the influence of Hurricane Earl (2010). This code is easily adaptable to other storm events using buoy data from the same source and adjusted dates.

---

Data Sources

• Oceanographic buoy data: NOAA Buoy Data Center

---

Dependencies

Install the following R packages to run the analysis:

install.packages(c("dplyr", "lubridate", "FactoMineR", "factoextra", "MASS"))

Usage

• Clone or download the repository.
• Ensure the bftn7h2010.txt file is in the working directory.
• Knit the .Rmd file to an HTML or PDF report for a full analysis.
• Data can be changed out for other buoy data .txt files provided the headers match. For each change in storm, the selected weeks before/after each storm must be changed in the code to account for the new storm date (lines 81-96).

Steps

1. Data Preparation

• Load and filter buoy data for valid environmental readings.
• Assign weeks to the dataset and filter periods before and after Hurricane Earl.

2. Dimensionality Reduction (PCA)

• Perform PCA on the buoy data to reduce variables.

• Generate visualizations for PCA results (e.g., biplots and scree plots).

• Analyze data separately for periods before and after the hurricane.

  Note

  The code for GLM at the end of the .rmd and the bridgenet.csv are part of a continuation to this project that has not yet been completed and can be disregarded at the moment.