Viet and Luke
CISC-820: Quantitative Foundations
Project 1: Linear Feature Engineering
The easiest way to install the dependencies is to use conda,
conda init
conda create --name py310 python=3.10.4
conda activate py310
conda install black numpy pytest scikit-learn
conda install -c pytorch pytorch
Alternatively, using Python 3.10.4, with pip and env,
python -m venv env
source env/bin/activate
pip install black numpy pytest scikit-learn torch==1.12.1
For the TA grading the assignment, this is likely the only command you will need to run after setting up the environment. To reproduce the results produced for submission, run python main.py --submission and the reuslt will be in the file testoutputs_nn.txt.
To run a bulk set of experiments, run
python main.py
The settings of the experiments can be tweaked with command line arguments, such as
python main.py -k 10 -p 3 -r results.json -s -v
A brief description of the flags can be found by running
python main.py --help
For developers, the unit tests are run with
pytest main.py
and the auto-formatter with
black *.py