1.8 KiB
Neural Networks for Breast Cancer Detection on Limited Featuresets
Lars Bogner at the University of Oslo
Project 2 for FYS-STK4155
Abstract
We study an approach to classification on the Wisconsin Breast Cancer Diagnostic dataset using neural networks to expand a limited set of physical features to the full feature set. Using only the radius and area of cell nuclei, we train a feedforward neural network to predict the remaining features with a mean squared error of 3.15 σ². Using out-of-fold prediction we obtain independent predictions for the entire dataset. Using these predictions as input for a logistic regression classifier, we achieve an accuracy of 94.74 % and an AUC score of 0.99105, comparable to results using the full dataset.
Installation
To run the code in this repository, use the package manager uv for the most seamless experience. Make sure you have uv installed on your system. You can find installation instructions at the project's homepage.
Then, navigate to the root directory of this repository in your terminal and run the following command to install all necessary dependencies:
uv install
Alternatively, you can manually install the required packages using pip. The main dependencies are listed in the requirements.txt file. You can install them by running:
pip install -r requirements.txt
You will also require the pyoptim library, which can be found at UIO GitHub.
Usage
To create the results and figures presented run all the notebooks in the notebooks/ directory. Make sure to run them in the correct order as some notebooks depend on the outputs of others. To compile the report, navigate to the report/ directory and run:
pdflatex main.tex