Predicting Heart Attack Risk Using Machine Learning

A machine-learning study that predicts heart-failure mortality risk from clinical patient data. A logistic-regression pipeline trained on 299 heart-failure records reaches 86.7% test accuracy across 13 clinical features (age, ejection fraction, serum creatinine, and more).

Heart Attack Failure Prediction - Jupyter Notebook Documentation

This repository contains a Jupyter Notebook for analyzing a dataset on heart attack prediction. The notebook explores the data, performs feature engineering, trains a machine learning model, and evaluates it's performance.

Data Description:

The dataset, "Heart Failure prediction.csv", contains 299 rows and 13 columns related to heart failure patients. The columns include:

• Demographic information: age, sex
• Medical history: anaemia, diabetes, high_blood_pressure, smoking
• Lab results: creatinine_phosphokinase, ejection_fraction, platelets, serum_creatinine, serum_sodium
• Target variable: DEATH_EVENT (1: death event, 0: no death event)

Notebook Content:

1. Imports:

   • Standard libraries: pandas, numpy, matplotlib, seaborn
   • Machine learning libraries: sci-kit-learn for data preprocessing, model training, and evaluation

2. Data Loading:

   • Loads the CSV data using pandas.read_csv()

3. Exploratory Data Analysis (EDA):

   • Provides basic summary statistics using describe().T
   • Analyzes data distribution and missing values
   • Creates visualizations (count plots, heatmaps) to understand relationships between features and the target variable (DEATH_EVENT)

4. Data Preprocessing:

   • Identifies categorical and numerical features
   • Encodes categorical features (if necessary)
   • Handles missing values (if any)
   • Splits the data into training and testing sets using train_test_split()

5. Model Training:

   • Uses Logistic Regression for heart attack prediction
   • Applies StandardScaler() for feature scaling

6. Model Evaluation:

   • Trains the model on the training set using fit()
   • Makes predictions on the testing set using predict()
   • Evaluates model performance using:
     • Classification report (precision, recall, F1-score)
     • Accuracy score
   • Compares training and testing set performance

Getting Started:

1. Clone this repository.
2. Install required libraries using pip install pandas numpy matplotlib seaborn scikit-learn
3. Open the Jupyter Notebook (e.g., Heart_Attack_Failure_Prediction.ipynb)
4. Run the code cells sequentially.