This project applies and compares the performance of the more classical supervised clasification method Random Forest and that of a Deep Neural Network to the task of spam classifications. The main finding is that the less computationally expensive model of the Random Forest achieves a better performance in terms of F1-score and accuracy while additionally delivering further model insights such as feature importance.
All plots created saved as png-files.
Model creation within a quarto-file. Folder also contains the rendered html.
Final report as a quarto and a rendered PDF file.
Descriptive statistics table stored as a tex-file.
The data source used for model training.