adekunle-feranmi/Kwasu-Complaint-System-

This is a demo for kwasu complaint system # KWASU Automated Complaint Management System A web-based system for Kwara State University that lets students submit complaints online, automatically classifies them with machine learning, flags abusive language for human review, and gives administrators a dashboard to verify students and manage complaints. > **Categories:** This build classifies complaints into **two** categories — > **Academic** and **Administrative** — because the provided dataset only > contains those two labels. (The original brief listed four; Facility and > Finance are not present in the data and cannot be predicted. See > `ml_training/training_results.txt`.) --- ## Tech stack | Layer | Technology | |--------------|----------------------------------------------| | Frontend | React (Vite) | | Backend | Python + Flask (REST API) | | Database | MySQL (SQLAlchemy ORM) | | ML | scikit-learn (Naïve Bayes + SVM, TF-IDF) | | Auth | JWT (login with email **or** matric number) | | Hosting | Vercel (frontend) · Render (backend) · Railway/PlanetScale (MySQL) | --- ## Features **Students** - Sign up with email + matric number + password; log in with **either** email or matric number - Complete a profile (name, matric, department, level) that an admin must verify - Editing the profile resets it to *pending* and requires re-approval - View the complaint feed; verified students can submit complaints and comment - Track personal complaint status and admin responses **Automatic processing on submit** 1. Abuse detection runs first (profanity wordlist + regex). If flagged → flagged queue, **not** published, **not** classified. 2. If clean → preprocessing → TF-IDF → chi-square feature selection → ML classification → stored as *Pending* and published to the feed. **Admins** (two dashboard sections) - **Profile Management:** approve / reject (with rea …