CIS-7 · Cosmic Intelligence System

Galactic Classification Challenge · Planet Survival AI

---

STEP 1 — Place your files

Copy these 3 files from your GitHub repo into this folder structure:

CIS_FINAL/
│
├── ml/
│   ├── Model/
│   │   ├── best_svc_Model.pkl        ← COPY HERE from GitHub
│   │   └── num_pipeline.pkl          ← COPY HERE from GitHub
│   │
│   ├── Dataset/
│   │   └── thermoracleTrain.csv      ← COPY HERE from GitHub
│   │
│   ├── Code/
│   │   └── app.ipynb                 (optional)
│   └── Predictions/
│       └── Final_predictions.csv     (optional)
│
└── api/
    ├── main.py       ← backend (do not move)
    ├── index.html    ← frontend (do not move)
    └── requirements.txt

Your GitHub repo path:

• Cosmic-Classifier/ml/Model/best_svc_Model.pkl
• Cosmic-Classifier/ml/Model/num_pipeline.pkl
• Cosmic-Classifier/ml/Dataset/thermoracleTrain.csv

---

STEP 2 — Run

Windows

Double-click START_WINDOWS.bat

Mac / Linux

chmod +x START_MAC_LINUX.sh
./START_MAC_LINUX.sh

Manual (any OS)

cd CIS_FINAL/api
pip install -r requirements.txt
uvicorn main:app --reload --port 8000

---

STEP 3 — Open browser

Go to: http://localhost:8000

⚠️ Do NOT open index.html directly as a file — it must go through the server.

---

What happens when you run it

Step	What the system does
Boot screen	Connects to backend, streams live log
0-5 sec	Loads your num_pipeline.pkl + best_svc_Model.pkl
5-10 sec	Reads thermoracleTrain.csv (60,000 rows)
10-60 sec	Trains 9 other models in background (LR, DT, RF, KNN, XGB, GB, ADA, NB, ET)
Live	Model tabs unlock one-by-one as each finishes, showing real accuracy
Ready	All 10 models available, galaxy scene running, planet morphs on classification

---

Troubleshooting

"Module not found: xgboost"

pip install xgboost

"Pipeline not loaded" error Make sure num_pipeline.pkl is in ml/Model/ not anywhere else.

Models tab shows "…" forever Backend is not running. Check the terminal — look for errors.

Page is blank / galaxy not showing Open DevTools (F12) → Console. If you see a Three.js error, try a different browser (Chrome works best).

Backend offline warning in UI The UI still works in simulation mode — just not using your real model. Start the backend with uvicorn main:app --reload --port 8000.

---

Tech Stack

Layer	Technology
Frontend	HTML + CSS + Three.js (r128)
Backend	FastAPI + Uvicorn
ML Pipeline	scikit-learn, XGBoost
Models	SVC, RF, XGB, GB, KNN, DT, LR, ADA, NB, ET
Dataset	thermoracleTrain.csv — 60,000 rows, 10 features
Best Model	SVC — 92% accuracy (GridSearchCV tuned)

---

Built for the Galactic Classification Challenge (GCC) · 2547