https://github.com/hetuvpatel/brain-stroke-prediction
Machine Learning project for predicting stroke risk using healthcare data. Includes EDA, preprocessing, SMOTE, feature selection (RFE), evaluation of Logistic Regression, Decision Tree, Random Forest, KNN, SVM, and Stacked Ensemble models.
https://github.com/hetuvpatel/brain-stroke-prediction
data-mining ensemble-learning healthcare machine-learning predictive-modeling python rfe scikit-learn smote
Last synced: 5 days ago
JSON representation
Machine Learning project for predicting stroke risk using healthcare data. Includes EDA, preprocessing, SMOTE, feature selection (RFE), evaluation of Logistic Regression, Decision Tree, Random Forest, KNN, SVM, and Stacked Ensemble models.
- Host: GitHub
- URL: https://github.com/hetuvpatel/brain-stroke-prediction
- Owner: hetuvpatel
- Created: 2025-04-27T20:31:17.000Z (5 days ago)
- Default Branch: main
- Last Pushed: 2025-04-28T07:47:27.000Z (5 days ago)
- Last Synced: 2025-04-28T08:36:33.061Z (5 days ago)
- Topics: data-mining, ensemble-learning, healthcare, machine-learning, predictive-modeling, python, rfe, scikit-learn, smote
- Language: Python
- Homepage: https://colab.research.google.com/drive/1jgS5OhntiZSgC0nOAxXHaTfI-8YK6H8t?usp=sharing
- Size: 1010 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# 🧠 Stroke Prediction Using Data Mining
[](https://www.python.org/)
[](https://scikit-learn.org/)
[](https://colab.research.google.com/)
[](https://opensource.org/licenses/MIT)
---
## 📖 Project Overview
**Stroke Prediction Using Data Mining** is a machine learning project that aims to build a predictive model to classify whether an individual is likely to suffer a stroke based on their healthcare and demographic data.
This project involved:
- Comprehensive data preprocessing and EDA
- Implementation of **5 ML models** (Logistic Regression, Decision Tree, Random Forest, K-Nearest Neighbors, SVM)
- Usage of techniques like **SMOTE** (for class imbalance) and **Recursive Feature Elimination (RFE)**
- Creation of an advanced **Stacked Ensemble Model** for improved predictive accuracy
---
## 📊 Dataset
- Source: [Kaggle - Stroke Prediction Dataset](https://www.kaggle.com/datasets/fedesoriano/stroke-prediction-dataset)
- Records: **5,110** instances
- Features:
- **Categorical**: Gender, Ever Married, Work Type, Residence Type, Smoking Status
- **Numerical**: Age, Hypertension, Heart Disease, Avg Glucose Level, BMI
*Note: BMI missing values were handled via mean imputation.*
---
## 🛠️ Techniques Used
| Technique | Purpose |
|:---------:|:-------:|
| **Label Encoding** | Convert categorical variables |
| **SMOTE** | Handle class imbalance |
| **Standard Scaler** | Normalize numerical features |
| **Recursive Feature Elimination (RFE)** | Feature selection |
| **Ensemble Learning** | Boost accuracy with multiple models |
---
## 📈 Models Evaluated
| Model | Accuracy | F1 Score | ROC AUC |
|:------|:--------:|:--------:|:-------:|
| Logistic Regression | 0.79 | 0.80 | 0.85 |
| Decision Tree | 0.91 | 0.91 | 0.91 |
| Random Forest | **0.96** | **0.96** | **0.99** |
| K-Nearest Neighbors (KNN) | 0.90 | 0.90 | 0.95 |
| Support Vector Machines (SVM) | 0.84 | 0.85 | 0.91 |
🚀 **Best Model**: Random Forest
🚀 **Even Better**: Stacked Ensemble achieved **96.66% accuracy**!
---
## 📊 Additional Observations
- **Tuned SVM** improved accuracy and ROC AUC through hyperparameter optimization.
- **Stacked Ensemble** combined base models and gave the best generalization.
---
---
## 🔥 Key Takeaways
- **Data Preprocessing** and **Feature Engineering** significantly affect performance.
- **SMOTE** improved stroke class detection.
- **Ensemble models** (Random Forest, Stacked) outperform individual models.
- **Feature Selection (RFE)** simplifies models with minor performance trade-offs.
---
## 📚 References
- 📂 Dataset: [Stroke Prediction Dataset - Kaggle](https://www.kaggle.com/datasets/fedesoriano/stroke-prediction-dataset)
- 📑 Project Notebook: [Google Colab Notebook](https://colab.research.google.com/drive/1jgS5OhntiZSgC0nOAxXHaTfI-8YK6H8t?usp=sharing)
---
## 📬 Contact
- 👤 Hetu Virajkumar Patel | [GitHub](https://github.com/hetuvpatel) | [LinkedIn](https://www.linkedin.com/in/hetu-patel-toronto/)
- 👤 Nilay Thakorbhai Patel
---
> 🎯 **Developed as part of CPS 844 - Data Mining, under Prof. Cherie Ding at Toronto Metropolitan University.**
---