🧠 Bankruptcy Prediction using Machine Learning

📘 Overview

This project predicts company bankruptcy using advanced machine learning models.
It demonstrates a complete end-to-end workflow — from data wrangling and resampling to model tuning, evaluation, and deployment — comparing three approaches: Decision Tree, Random Forest, and Gradient Boosting.

The goal is to identify financial risk early by analyzing company metrics such as liquidity, leverage, and profitability indicators.

---

⚙️ Project Highlights

• 🧩 Data Preprocessing: Handled missing values with SimpleImputer and balanced the dataset using RandomOverSampler.
• 🌳 Decision Tree: Built an interpretable baseline model.
• 🌲 Random Forest: Improved model generalization through ensembling.
• 🚀 Gradient Boosting: Delivered the highest accuracy by sequentially correcting model errors.
• 📊 Evaluation: Used cross-validation, accuracy, precision, recall, and confusion matrices for model assessment.
• 💾 Deployment: Serialized the best model (model-5-3.pkl) and implemented a reusable make_predictions() function.

---

🔧 Hyperparameter Tuning

Model optimization was performed using GridSearchCV with 5-fold cross-validation to ensure robustness and prevent overfitting.
Key hyperparameters tuned:

• imputer__strategy: ["mean", "median"]
• max_depth: [10, 20, 30, 40, 50]
• n_estimators: [25, 50, 75, 100]

This systematic tuning enhanced performance and reduced bias across all ensemble models.

---

🧰 Tech Stack

Python, pandas, scikit-learn, imbalanced-learn, matplotlib, ipywidgets

---

📈 Results

Model	Accuracy	Highlights
Decision Tree	Moderate	Baseline interpretability
Random Forest	Higher	Reduced overfitting
Gradient Boosting	Best	Balanced precision & recall

---

🙋‍♂️ About Me

Hi, I’m Pushkin Kumar — a data enthusiast passionate about building predictive models and turning data into actionable insights.

📫 LinkedIn | GitHub

---