spam_detection

Purpose

The aim of this project is to recognize spam users among Decidim.

How does it work:

A decision tree classifier has been trained to assign a probability to every user of a decidim database.

Getting Started

Requirements

• Python >= 3.8
• pip >= 21.x.x

Note:

This project can be executed only by decidim admins, the data has to be passed by the HTTP method: request

Deployment

To deploy this project run:

  pip install -r requiremts.txt
  python app.py

Training

1. Put a csv inside the data folder with the following columns: 'sign_in_count', 'personal_url', 'about', 'avatar', 'extended_data', 'followers_count', 'following_count', 'invitations_count', 'failed_attempts', 'admin', 'is_spam' You can use train_example.csv as example.
2. make train
3. Move new_model.pkl and replace model.pkl in the main folder.