Mood Swing Prediction API and Streamlit Web App

Problem Statement

In today's fast-paced world, mental health challenges like mood swings often go unnoticed or undiagnosed, leading to severe consequences for individuals' well-being. While professional care is essential, early detection using accessible and non-invasive methods can help people become more aware of their mental state and seek support when necessary. This project aims to provide a lightweight, AI-powered solution that predicts mood swing tendencies based on survey-style inputs, making it easier to monitor and reflect on mental health.

Project Overview

• A FastAPI-based RESTful API for making predictions.
• A Streamlit web application for users to interact with the model via a simple UI.

Features

• Input form for mental health and lifestyle-related data
• Real-time mood swing prediction
• Simple API to connect front-end and back-end

Tech Stack

• Python 3.10+
• Scikit-learn
• XGBoost
• Pandas, NumPy
• FastAPI
• Streamlit
• Joblib

Folder Structure

├── app/
│   ├── main.py             # FastAPI app for serving predictions
├── model/
│   ├── xgb_model.pkl       # Trained XGBoost model
│   ├── encoder.pkl         # Saved label encoders (as a dictionary)
│   ├── train_model.py      # Script to train and save the model
├── streamlit_app/
│   ├── app.py              # Streamlit web application
├── data/
│   ├── raw_data.csv         # Original dataset
│   └── cleaned_data.csv     # Cleaned and preprocessed dataset
├── PipFile                  # Python dependencies
└── README.md

MoodFlag Project Resources

Frontend

• Live App:
  MoodFlag Streamlit App

Backend (API)

• API Docs (Swagger UI):
  📘 MoodFlag API Documentation

Dataset & Model Exploration

• Kaggle Notebook:
  📈 ML Applications to Mental Health Diagnosis

---