The folder with the backend module. Uses Flask. See the README inside the folder for more information.

The folder with the frontend module. Uses React. See the README inside the folder for more information.

• EDA.ipynb: Exploratory Data Analysis Notebook which explores the machine learning process to reach the final model.
• model_trainer.py: The final model generator. Running the file will reproduce a file called rf_model.pkl which is the model produced by the script, as well as output the testing and evaluation.
• requirements.txt: The requirements used in both the files mentioned above.

The model is produced using a Random Forest Classifier with the following hyperparameters:

• Max Depth: 4
• # of Estimators: 100

The choice of the each hyperparameter was determined through experimentation, with any depth greater than 4 did not provide any additional benefit, while the number of estimators was chosen as 100 as it adds in sufficient smoothing without underfitting overall.

Create a virtual environment with the requirements installed first.

$ virtualenv --python=`which python3` venv
$ source ./venv/bin/activate
$ pip install -r requirements.txt

Run the model trainer:

$ python model_trainer.py

A file rf_model.pkl will be generated.

This model can be reloaded by using Pickle to load the binary and predictions can be made.