🧠 Stroke Prediction Using PySpark Machine Learning

📌 Project Overview

This project develops a scalable machine learning pipeline using PySpark to predict the likelihood of stroke occurrence based on patient demographic and clinical information.

The project demonstrates the complete machine learning workflow, including data preprocessing, feature engineering, dimensionality reduction, and model development using Logistic Regression. PySpark was used to efficiently process healthcare data and prepare it for predictive modeling.

---

🎯 Objectives

• Analyze healthcare data using PySpark.
• Clean and preprocess the dataset.
• Convert data into appropriate formats for machine learning.
• Perform feature engineering and dimensionality reduction.
• Build a Logistic Regression model for stroke prediction.
• Evaluate the effectiveness of the prediction model.

---

📊 Dataset

This project uses the Stroke Prediction Dataset from Kaggle.

Dataset Information

• Total Records: 5,110
• Features: 12
• Target Variable: Stroke

Features

Feature	Description
gender	Patient gender
age	Age
hypertension	Hypertension status
heart_disease	Heart disease status
ever_married	Marital status
work_type	Employment type
Residence_type	Urban/Rural
avg_glucose_level	Average glucose level
bmi	Body Mass Index
smoking_status	Smoking history
stroke	Target variable (0 = No Stroke, 1 = Stroke)

Dataset Source

https://www.kaggle.com/datasets/fedesoriano/stroke-prediction-dataset

---

🛠 Technologies Used

• Python
• PySpark
• Apache Spark
• Pandas
• NumPy
• Scikit-learn
• Matplotlib
• Seaborn

---

📂 Repository Structure

```text
stroke-prediction-using-pyspark/
│
├── README.md
├── requirements.txt
├── .gitignore
│
├── data/
│   ├── healthcare-dataset-stroke-data.csv
│   └── stroke_dataset.csv
│
├── src/
│   ├── data_preprocessing.py
│   ├── datatype_changing.py
│   ├── dimensionality_reduction.py
│   └── logistic_regression.py