This repository contains a collection of Machine Learning (ML), Deep Learning (DL), and Optimization projects, covering various domains and real-world applications. Each project explores different models, techniques, and libraries to solve complex problems efficiently.

For Git and GitHub setup:

• GitHub Repository Setup

/
├── README.md                       # Main documentation
├── docs/
│   ├── Github.md                    # Git and GitHub setup instructions
├── Bankruptcy_Prediction            # A project on predicting bankruptcy for companies
├── Healthcare - Length of Stay      # A project on predicting the length of stay of a patient
├── Cool Wipes - Linear Programming  # A new project on supply chain optimization
└── ...

• Diverse Domains: Healthcare, finance, e-commerce, and more.
• Model Categories:
  • Regression and Classification Models
  • Clustering and Dimensionality Reduction
  • Time Series Models (ARIMA, SARIMA, Prophet, etc.)
  • Deep Learning (CNNs, RNNs, Transformers, GANs)
  • Ensemble Models (Random Forest, XGBoost, LightGBM)
• Libraries and Tools: pandas, NumPy, Matplotlib, Seaborn, scikit-learn, TensorFlow, PyTorch, Keras, statsmodels, and more.
• Complete Workflows:
  • Problem definition
  • Data preprocessing
  • Feature engineering
  • Model development and evaluation
  • Insights and interpretation

Predicts the likelihood of bankruptcy for companies to aid in risk assessment and financial planning.

• Data: Financial and operational metrics of companies with 96 variables and 6819 records. The target variable indicates bankruptcy status.
• Methods: Logistic Regression, XGBoost, and Ensemble Methods with hyperparameter tuning for improved accuracy.
• Evaluation: Accuracy, Precision, Recall, F1 Score, and model interpretation using ensemble predictions.

Project Directory: Bankruptcy Prediction

Predicts hospital patient length of stay (LOS) to enhance resource allocation and care management.

• Data: Patient and hospital records categorized into 11 LOS classes.
• Methods: Logistic Regression, Random Forest, Gradient Boosting, LightGBM, CatBoost, SVM, XGBoost.
• Evaluation: Accuracy, Precision, Recall, F1 Score, ROC AUC.

Project Directory: Patient Length of Stay Prediction

Optimizing the production and distribution network for CoolWipes using Gurobi to minimize costs while meeting demand across six geographic regions.

• Data: Demand data for wipes and ointments across six regions, current plant capacities, fixed and variable costs, and transportation costs.
• Methods:
  • Gurobi Optimization: Formulating a linear programming (LP) model to determine the optimal production and distribution strategy.
  • Scenario Analysis: Evaluating cost structures under different transportation cost assumptions.
• Evaluation:
  • Baseline Analysis: Assessing the annual cost of serving the entire nation from Chicago.
  • Expansion Decision: Evaluating the impact of adding new plants in Princeton, Atlanta, or Los Angeles.
  • Optimal Network Design: Recommending the best plant locations and capacities under varying constraints.
  • Future Planning: Projecting network structure for 2026 with a 35% demand increase and next-day delivery.

• What is the cost of serving the nation from a single plant in Chicago?
• Should additional plants be built? If so, where and with what capacity?
• How does transportation cost variability influence plant location decisions?
• What is the best network design if starting from scratch?
• Can the network support projected demand growth by 2026?
• How can AI and smart technologies improve supply chain efficiency?

Project Directory: Cool Wipes - Linear Programming

Contributions are welcome! Fork the repository, make your changes, and submit a pull request. For major changes, open an issue for discussion.