Overview of simple Random Forest vs Suport Vector Machine.

DISCLAIMER: This script is for demonstrative propouses only and should not be used as is in production.

Objective

The propose of this repository is to showcase the implementation of code for text classification methods. Hyperparameter tunning, data augmentation and other methods are not in the scope.

In the code are mentioned decisions made through the process but they do not dig into the statistical or mathematical backgrounds.

Requirements

This project requires Python <= 3.12. I recommed to make a virtual enviorment to test it.

Test a Model Release

1. Clone the repo

git clone https://github.com/hriva/random-forest-v-svm.git
cd random-forest-v-svm

# Make virt-env
python3 -m venv
source .venv/bin/activate
pip install -r requirements.txt

2. Download one of the pre-trained models

3. Run the test_mail.ipynb Notebook

Train the Model with Notebook

1. Clone the repo

git clone https://github.com/hriva/random-forest-v-svm.git
cd random-forest-v-svm

# Make virt-env
python3 -m venv
source .venv/bin/activate
pip install -r requirements.txt

2. Download the data set (not in this repo) to the working directory.

Yout need a kaggle account to download the data set. https://www.kaggle.com/datasets/subhajournal/phishingemails/download?datasetVersionNumber=1

4. Unzip the dataset.

# this creates a directory and extracts there.
unzip archive.zip -d data

5. Run the notebook.

Train the Model with Web Interface

1. Clone the repo

git clone https://github.com/hriva/random-forest-v-svm.git
cd random-forest-v-svm

# Make virt-env
python3 -m venv
source .venv/bin/activate
pip install -r requirements.txt

2. Start the web Interface

streamlit run app.py

3. Download the data set (not in this repo) to the working directory.

Yout need a kaggle account to download the data set. https://www.kaggle.com/datasets/subhajournal/phishingemails/download?datasetVersionNumber=1

4. Unzip the dataset.

# this creates a directory and extracts there.
unzip archive.zip -d data

5. Open the web app url http://localhost:8501/
6. Select the csv to train the model. After the model is trained a text input will appear.
7. Enter the body of the mail you want to test in the text input.