afcatPYQ — AFCAT 2026 Prediction & Dashboard

A data science pipeline that analyzes previous-year questions (PYQ) to predict high-priority topics, generate study plans, and publish an interactive dashboard for AFCAT 2026 preparation.

⚠️ Disclaimer: This system predicts topic patterns and distributions (60–75% accuracy), not exact questions (25–40% accuracy at best). Use it to optimize preparation, not as a crystal ball.

---

Features

• Topic Frequency Analysis — Identifies high-probability topics from historical data
• Trend Detection — Tracks rising and falling topic importance across years
• Difficulty Prediction — Classifies expected difficulty levels per topic
• Current Affairs Scoring — Rates news articles by AFCAT relevance
• Study Plan Generator — Produces an optimized 20-day study plan and mock-test blueprint
• AI Practice Questions — Generates practice questions based on predicted patterns
• Static Dashboard — Publishes an HTML + JSON dashboard under output/predictions_2026/

---

Project Structure

afcatPYQ/
├── data/
│   ├── raw/                  # Raw question papers (JSON/CSV)
│   ├── processed/            # Cleaned and labeled data
│   └── sample/               # Sample data for testing
├── models/
│   ├── topic_predictor.py    # XGBoost topic frequency model
│   ├── difficulty_model.py   # Random Forest difficulty classifier
│   └── current_affairs.py   # News relevance classifier
├── utils/
│   ├── data_loader.py        # Data loading and preprocessing
│   ├── feature_engine.py     # Feature engineering
│   └── bias_correction.py   # Memory recall bias correction
├── analysis/
│   ├── topic_analyzer.py     # Historical topic analysis
│   └── trend_detector.py    # Temporal trend detection
├── scripts/
│   ├── generate_complete_data.py  # Dashboard bundle generator
│   └── mark_certain_topics.py    # Topic confidence marker
├── output/
│   └── predictions_2026/    # Generated dashboard (data.js, JSONs, dashboard.html)
├── .github/workflows/
│   └── deploy-dashboard.yml # CI/CD for GitHub Pages
├── main.py                  # Main entry point
├── config.py                # Configuration settings
└── requirements.txt         # Dependencies

---

Quick Start

# 1. Install dependencies
pip install -r requirements.txt

# 2. Run the prediction system
python main.py

# 3. Generate a study plan
python main.py --generate-plan

# 4. Analyze current affairs
python main.py --analyze-news

# 5. Regenerate the dashboard bundle
python scripts/generate_complete_data.py

# 6. Preview the dashboard locally
python -m http.server 8000 --directory output/predictions_2026
# Open http://localhost:8000/dashboard.html

---

Deployment

Vercel (Recommended)

1. Rename the dashboard entry point before pushing:

   mv output/predictions_2026/dashboard.html output/predictions_2026/index.html
   git add output/predictions_2026/index.html
   git commit -m "chore: rename dashboard to index.html for Vercel"
   git push
   

2. In Vercel, import your GitHub repo and set:
   • Framework Preset: Other
   • Build Command: (leave empty, or python scripts/generate_complete_data.py to build on deploy)
   • Output Directory: output/predictions_2026
3. Deploy. Vercel will serve index.html alongside JSON files, so fetch('afcat_2026_predictions.json') works without path changes.

GitHub Pages

Use the included .github/workflows/deploy-dashboard.yml — it regenerates the dashboard and publishes to GitHub Pages automatically. You can also use gh-pages branch or the actions-gh-pages action.

Netlify

Set the build command to python scripts/generate_complete_data.py and the publish directory to output/predictions_2026.

---

Expected Accuracy

Prediction Type	Accuracy	Best Used For
Topic Distribution	65–75%	Knowing which topics appear
Question Count per Topic	±1–2 questions	Time allocation
Difficulty Level	55–60%	Preparation intensity
Specific Questions	25–40%	Identifying similar patterns

---

Key Insights for AFCAT 2026

High-Priority Topics (80%+ Confidence)

Subject	Topics
Math	Speed/Distance, Time/Work, Profit/Loss
English	Cloze Test, OLQ Vocabulary (Synonyms/Antonyms)
Reasoning	Dot Situation, Spatial Ability, Venn Diagrams
GK	Defence Operations, Sports Terms, Modern History

Structural Changes for 2026

• Math section increased from 18 → 20 questions
• EKT discontinued (may shift weight to GK science questions)
• Greater focus on Officer-Like Qualities (OLQ) vocabulary

---

Configuration

Edit config.py to customize:

• Historical data years to include
• Confidence thresholds
• Topic categories
• Model hyperparameters

---

Development Notes

• pyqData is embedded in data.js during bundle generation. The dashboard script copies top-level consts into window.* so the UI works when data.js defines const pyqData = ... rather than window.pyqData.
• To debug topic-to-PYQ links, open the browser console and inspect window.pyqData and window.predictionsData.
• Large PDF scans and data files may exist under data/. Consider removing binary artifacts from the repo and storing them externally if size becomes an issue.

---

License

For educational purposes only. Use legally acquired data sources.

---

Contributing / Support

Open an issue on GitHub for help with automated deploys to Vercel, Netlify, or GitHub Pages.