Author: Bradley Reardon
Code can be found at https://github.com/bradreardon/cs4641-supervised-learning.
- Only tested with Python 3.7+ (though lower in-support versions of Python 3 may be sufficient)
- Packages specified in
requirements.txt
:- Install with
pip install -r requirements.txt
- Install with
Before running, make sure that the out
directory exists in the same place as main.py
has the following structure:
out
|- boosting
|- decision_tree_pruning
|- knn
|- neural_net
|- svm
This can be done with the command mkdir -p out/{boosting,decision_tree_pruning,knn,neural_net,svm}
.
main.py
provides usage instructions. In general, commands follow the format python main.py <algorithm>
.
For some algorithms, parameters may be specified on the CLI, though some of these cases are unimplemented.
Running each script will output statistics (and warnings!) for the training of the algorithm, and will export charts
to subdirectories of the out
directory.
scikit-learn provided excellent documentation, and some of the charts generated by this program can be found in the scikit-learn documentation. Namely, the boosting error chart was modified from the documentation online.