Module 8: Model Deployment
• Overview of model deployment techniques
• - Saving models (Pickle/Joblib)
• - Flask/Django
• - Streamlit & FastAPI
• - Cloud Deployment
• - Project: House Price Prediction App
Saving ML/DL Models
• Why save models?
• - Avoid retraining
• - Reusability
• - Deployment ready
Pickle - Saving Models
• Code:
• import pickle
• [Link](model, open('[Link]','wb'))
• Load:
• model = [Link](open('[Link]','rb'))
Pickle Example Output
• Input: [5]
• Output: [10.0]
Joblib - Saving Models
• Code:
• import joblib
• [Link](model,'[Link]')
• Load:
• model = [Link]('[Link]')
Joblib Example Output
• Input: [7]
• Output: [14.0]
Flask for Model Deployment
• Simple Flask API for prediction
• - Lightweight
• - Good for quick APIs
Flask Example Code
• POST /predict
• Input: {"features":[5]}
• Output: {"prediction": [10.0]}
Django for Model Deployment
• Best for larger apps
• - Authentication
• - Database integration
• - Admin interface
Streamlit for Deployment
• Easiest UI framework
• - Python-only
• - Fast prototyping
• - Interactive inputs
Streamlit Example Code
• [Link]('House Price Prediction')
• sqft = st.number_input('Enter sqft')
• [Link]([[sqft]])
FastAPI for Deployment
• Modern, fast API
• - Async support
• - Easy Swagger docs
FastAPI Example Code
• POST /predict
• Input: [1200,3,2]
• Output: {'prediction':[250000]}
Cloud Deployment Basics
• Platforms:
• - AWS (EC2, SageMaker)
• - GCP (App Engine, AI Platform)
• - Azure (App Service, ML Studio)
Deployment Workflow
• Steps:
• 1. Train & Save Model
• 2. Build Flask/Streamlit App
• 3. Containerize with Docker
• 4. Deploy to Cloud
Project: House Price Prediction
• Steps:
• 1. Train regression model
• 2. Save with joblib
• 3. Build Streamlit UI
• 4. Deploy on Cloud
Project Input Example
• User enters:
• - Square feet
• - Bedrooms
• - Bathrooms
Project Output Example
• Prediction: $350,000
Deployment Options
• Local Deployment: streamlit run [Link]
• Cloud Deployment: AWS/GCP/Azure/Heroku
Conclusion
• You learned:
• - Saving models
• - Flask, Django, Streamlit, FastAPI
• - Cloud basics
• - Built a real deployment project
