Heart disease analysis and prediction [XGBoost, SVM, KNN, LR]

Description

Analysis and prediction of heart disease using various machine learning models.

Motivation

According to WHO, cardiovascular diseases are the leading cause of death globally, taking an estimated 17.9 million lives each year. Identifying those at highest risk of CVDs and ensuring they receive appropriate treatment can prevent premature deaths [1]

Data

Dataset contains medical data of patients from Cleveland, US with suspected or diagnosed heart disease.

Authors

Janosi A., Steinbrunn W., Pfisterer M., & Detrano R. (1988). Heart Disease. UCI Machine Learning Repository.

Data description

Variable Name	Role	Type	Description	Units	Missing Values
age	Feature	Integer	Age	years	no
sex	Feature	Categorical	Sex		no
cp	Feature	Categorical	Chest pain type		no
trestbps	Feature	Integer	Resting blood pressure (on admission to the hospital)	mm Hg	no
chol	Feature	Integer	Serum cholestoral	mg/dl	no
fbs	Feature	Categorical	Fasting blood sugar > 120 mg/dl		no
restecg	Feature	Categorical	Resting electrocardiographic results		no
thalach	Feature	Integer	Maximum heart rate achieved		no
exang	Feature	Categorical	Exercise induced angina		no
oldpeak	Feature	Integer	ST depression induced by exercise relative to rest		no
slope	Feature	Categorical	The slope of the peak exercise ST segment		no
ca	Feature	Integer	Number of major vessels (0-3) colored by flourosopy		yes
thal	Feature	Categorical	Thalium stress test result		yes
num	Target	Integer	Diagnosis of heart disease		no

Notes

• Dataset has missing values
• The num field refers to the presence of heart disease in the patient.
• Experiments with the Cleveland database have concentrated on simply attempting to distinguish presence (values 1,2,3,4) from absence (value 0).
• Dataset contains 303 instances and 14 attributes

Code

Data fetching

Data is fetched from the UCI repository and saved as a .csv file.

Visualization

• Data exploration was done using pandas library.
• Using seaborn and matplotlib, I created the following visualizations:
  • correlation matrix to see the correlation between the features
    
  • distribution of the features
    
  • pairplot of the features
    

Preprocessing

• Target variable was transformed to binary variable (0 - no heart disease, 1 - heart disease)
• Outliers detection was done using IQR method. Outliers were transformed to the upper and lower bounds.

Model training and evaluation

Models

• The data were split into training and testing sets.
• The following models were used:
  • XGBoost
  • Support Vector Machine
  • K Nearest Neighbors
  • Logistic Regression
• A pipeline was created for each model to scale the data and train the model.
• Cross-validation was employed to evaluate the models.
• Cross-validation was particularly utilized due to the small dataset size to avoid overfitting.
• The following metrics were utilized to evaluate the models:
  • Accuracy
  • Precision
  • Recall

Results

• Results were dependent on the random state of the train-test split and cross-validation.
• XGBoost, Supported Vector Machine and Logistic Regression models performed equally well.
• Models performed better than their baseline counterparts from UCI

Model	Average Performance	Baseline
XGBoost	0.84	0.81
SVM	0.84	0.65
Logistic Regression	0.83	0.81
KNNeighbors	0.82	N/A

References

[1] https://www.who.int/health-topics/cardiovascular-diseases