https://github.com/eshansugeesh/amex-machine-learning-project
Predictive modeling on American Express dataset using advanced ML algorithms, extensive feature engineering (300+ features), and cloud-based data science workflows.
https://github.com/eshansugeesh/amex-machine-learning-project
cross-validation customer-segmentation data-analytics data-science feature-engineering financial-data google-colab gradient-boosting hyperparameter-tuning lightgbm machine-learning neural-networks pandas predictive-modeling python reproducibility xgboost
Last synced: 2 months ago
JSON representation
Predictive modeling on American Express dataset using advanced ML algorithms, extensive feature engineering (300+ features), and cloud-based data science workflows.
- Host: GitHub
- URL: https://github.com/eshansugeesh/amex-machine-learning-project
- Owner: EshanSugeesh
- Created: 2025-10-01T14:51:07.000Z (10 months ago)
- Default Branch: main
- Last Pushed: 2025-10-01T14:57:32.000Z (10 months ago)
- Last Synced: 2025-10-01T16:36:05.882Z (10 months ago)
- Topics: cross-validation, customer-segmentation, data-analytics, data-science, feature-engineering, financial-data, google-colab, gradient-boosting, hyperparameter-tuning, lightgbm, machine-learning, neural-networks, pandas, predictive-modeling, python, reproducibility, xgboost
- Language: Jupyter Notebook
- Homepage:
- Size: 8.79 KB
- Stars: 0
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Amex Machine Learning Project
## Objective
Modeled customer behavior and transaction patterns for the American Express (Amex) dataset to predict financial outcomes and inform strategic decision making.
## Dataset & Setup
- Worked with complex, high-dimensional datasets provided in multiple formats (CSV, Parquet) containing event, offer, and transaction records
- Integrated Google Colab and Google Drive for scalable data access and secure storage
## Pipeline & Engineering
- Extensive feature engineering with over **300 handcrafted features** capturing spending patterns, account activity, offer interactions, and event-based customer actions
- Applied data cleaning, missing value imputation, and robust encoding strategies to maximize model interpretability and reduce noise
- Leveraged the Amex data dictionary for precise mapping and validation
## Model Development
- Utilized advanced machine learning algorithms (Gradient Boosting, XGBoost, LightGBM, and Neural Networks) for large-scale predictive modeling
- Performed iterative model selection, hyperparameter tuning, and cross-validation for optimal performance
- Built custom evaluation metrics aligned with competition objectives and business goals
## Submission & Reporting
- Generated prediction files in competition-ready format
- Codebase built for reproducibility: modular, well-documented workflow suitable for deployment and collaboration
## Impact
- Delivered actionable insights for customer segmentation, risk scoring, and offer targeting
- Demonstrated high proficiency in large-scale feature engineering, advanced ML pipelines, and cloud-based data science
- Project actively referenced in resume to highlight expertise in handling real-world, multi-source financial data and competitive ML modeling