title	Question Level Prediction
emoji	🎓
colorFrom	purple
colorTo	blue
sdk	docker
app_port	7860
app_file	app.py
python_version	3.9
pinned	false

🎓 Question Bloom Level & Difficulty Prediction

A machine learning project that classifies educational questions into Bloom's Taxonomy levels and Difficulty categories using Logistic Regression. The project implements a custom NLP pipeline with Sentence Transformers and provides a beautified Streamlit interface for real-time predictions.

---

📋 Table of Contents

• Overview
• Dataset
• Project Workflow
• Results
• Tech Stack
• Project Structure
• Getting Started

---

🔍 Overview

This project performs an end-to-end machine learning pipeline for analyzing the cognitive complexity and difficulty of educational questions. It covers:

1. Data Exploration — Analyzing question text distributions and student performance metadata
2. Feature Engineering — Generating semantic embeddings and normalizing student success metrics
3. Model Training & Evaluation — Developing calibrated Logistic Regression models for multi-class classification
4. Deployment — Creating a modern, interactive dashboard for production-ready inference

---

📊 Dataset

Note

This dataset was synthesized using a Large Language Model (LLM) due to the scarcity of publicly available datasets for multi-class Bloom's Taxonomy classification on specific educational content.

Property	Details
File	final.csv
Rows	5,500
Columns	~11 predictive columns
Target Variables	bloom_level, difficulty

Features

The model uses a total of 12 features (9 base features from the dataset + 3 engineered features):

Feature	Type	Source	Description
Question Text	string	Base	The raw text of the question (Vectorized via NLP)
Subject	object	Base	Broad subject category (e.g., Science, Maths)
Topic	object	Base	Specific topic within the subject
Avg Score	float64	Base	Average score achieved by students (0.0 - 1.0)
Correct %	float64	Base	Percentage of students who got the question right
Students Attempted	int64	Base	Total count of students who answered the question
Students Correct	int64	Base	Total count of students who answered correctly
Time Taken	float64	Base	Average time spent on the question (minutes)
Success Rate	float64	Engineered	Calculated as (Correct / Attempted)
Log Attempts	float64	Engineered	Log-transformed attempt count for better scaling
Question Length	int64	Engineered	Total word count of the question text

---

⚙️ Project Workflow

1. Data Cleaning & Engineering

• NLP Processing: Text is vectorized using SentenceTransformer ('all-MiniLM-L6-v2') to capture semantic intent.
• Categorical Encoding: One-Hot Encoding applied to Subject and Topic features.
• Scaling: Standardized numerical metrics using StandardScaler for model stability.

2. Implementation Approach

• Standalone Module: All logic encapsulated in BloomModelDeployer class for modular usage.
• Balanced Weights: Implemented class_weight='balanced' to handle imbalanced levels in Bloom's Taxonomy.

3. Training & Evaluation

• Splitting data into 80% training and 20% testing sets.
• Model Selection: During experimentation, XGBoost and Random Forest were tested. However, they did not provide a significant improvement in accuracy for this specific categorical text task, leading to the selection of Logistic Regression for its better generalization and interpretability.
• Persistence of all artifacts (models, encoders, scalers) into the models/ directory.

---

📈 Results

Metric (Accuracy)	Bloom Level	Difficulty
Accuracy Score	0.33	0.40
F1 Score (Macro)	0.33	0.39

Important

The current accuracy levels are primarily limited by the synthesized nature of the dataset. LLM-generated data, while useful for bootstrapping, often lacks the subtle nuances of real-world educational assessments, which affects the model's ability to reach higher precision.

---

🛠️ Tech Stack

Library	Purpose
Python 3.9	Programming language
Streamlit	UI Framework & Dashboard
Sentence-Transformers	NLP & Semantic Embeddings
scikit-learn	ML Models, Preprocessing, and Metrics
Pandas / NumPy	Data manipulation and Numerical logic
Docker	Containerization
Hugging Face	Deployment

---

📁 Project Structure

capstone_genai/
├── models/                    # Saved .pkl joblib artifacts
├── final.csv                  # Main dataset
├── milestone1.ipynb           # Research and Development Notebook
├── logistic_regression_deployment.py # Core ML class
├── app.py                     # Beautified Streamlit Frontend
├── requirements.txt           # Deployment dependencies
└── README.md                  # Project documentation

---

🚀 Getting Started

Prerequisites

• Python 3.9+
• Pip (Python Package Manager)

Installation

# Install dependencies
pip install -r requirements.txt

# Train the models (if pkl files are missing)
python3 logistic_regression_deployment.py --train

# Launch the dashboard
streamlit run app.py

Running with Docker

# Build the Docker image
docker build -t question-classifier .

# Run the container
docker run -p 7860:7860 question-classifier

---

📝 License

This project is part of the GenAI Capstone project.