Skip to content

README.md

Latest commit

 

History

History
86 lines (60 loc) · 3.09 KB

README.md

File metadata and controls

86 lines (60 loc) · 3.09 KB

AWS SageMaker Mobile Price Classification

Sagemaker test (mind that this is with older sagemaker v1 version)

Project Overview

This project builds a machine learning pipeline to classify mobile phone prices into different ranges using AWS SageMaker. The workflow includes:

  • Data Preparation: Load and split data using scikit-learn
  • Data Storage: Upload training/testing data to Amazon S3
  • Model Training: Train RandomForest classifier using SageMaker's SKLearn container
  • Model Deployment: Deploy trained model as a real-time inference endpoint
  • Predictions: Make predictions using the deployed endpoint
  • Cleanup: Delete endpoints to avoid unnecessary charges

Dataset

  • File: mob_price_classification_train.csv
  • Task: Multi-class classification (predicting price_range)
  • Features: 20 mobile phone specifications
  • Train/Test Split: 85% training, 15% testing

Requirements

  • AWS Account with SageMaker access
  • IAM role with SageMaker permissions
  • Python 3.x with dependencies in requirements.txt
  • S3 bucket for storing training data

WARNING (Testing only - IAM & Networking): Use least-privilege IAM roles and scoped policies; avoid granting broad/full access in production. Also avoid deploying resources into default/public subnets for production — prefer private subnets, NAT gateway or VPC endpoints, and properly scoped security groups.

Key Components

  1. research.ipynb - Main notebook containing the complete SageMaker workflow
  2. script.py - Training script executed by SageMaker
  3. train-V-1.csv - Training dataset
  4. test-V-1.csv - Testing dataset

Usage

Run the cells in research.ipynb to:

  1. Load and explore the dataset
  2. Split data into train/test sets
  3. Upload data to S3
  4. Train a RandomForest model on SageMaker
  5. Deploy the model to an endpoint
  6. Make predictions on test data
  7. Clean up resources

Model Performance

The RandomForest classifier is trained with:

  • n_estimators: 100
  • random_state: 0
  • Evaluation: Accuracy score and classification report on test data (15% of dataset)

🧹 Cleanup (Important: Avoid AWS Costs)

⚠️ To prevent unexpected charges, clean up these AWS SageMaker resources after testing:

  1. Delete SageMaker Endpoint

    • AWS Console → SageMaker → Endpoints → Select endpoint → Delete
    • This is critical as running endpoints incur hourly charges

  2. Delete Endpoint Configuration

    • AWS Console → SageMaker → Endpoint Configurations → Select config → Delete

  3. Delete Model

    • AWS Console → SageMaker → Models → Select model → Delete

  4. Clean S3 Bucket

    • AWS Console → S3 → Select bucket → Delete all training/test data and artifacts
    • Delete the bucket if created specifically for this project

  5. Remove IAM Role (if created specifically for this project)

    • AWS Console → IAM → Roles → Select role → Delete

  6. Delete CloudWatch Logs (optional)

    • AWS Console → CloudWatch → Logs → Delete log groups related to SageMaker

References

  • MLOPS Udemy course from Krish Naik