AWS-Sagemaker-Project

This project focuses on building a comprehensive data science pipeline to tackle a real-world business problem using machine learning. Our goal was to use Amazon SageMaker to develop a binary classification model that provides valuable insights and supports better decision-making. We started by clearly defining the problem and understanding why machine learning was the right solution. With a well-outlined objective, we framed the problem as a binary classification task and moved forward with the data analysis. The initial stages involved exploring and preparing the data, which included merging multiple CSV files, performing feature engineering, and handling any missing or inconsistent values. We ensured the dataset was ready for modeling by carefully encoding categorical variables and making adjustments as needed. Once the data was prepped, we trained and evaluated our model using a two-step approach. First, we built a baseline model to set a performance benchmark. Then, we experimented with more complex algorithms like XGBoost to enhance accuracy. Both models were trained and deployed on AWS to take advantage of SageMaker’s scalability and efficiency. Throughout the project, we prioritized making our code reproducible and portable, using relative paths and comprehensive documentation to make it easy for others to use. Finally, we created visualizations in Tableau to make the results more digestible. The dashboard highlighted key insights and performance metrics, giving a clear picture of the model’s effectiveness. The project wraps up with a reflection on the model's performance, the obstacles we faced, and the valuable lessons learned along the way.