Heart Failure Prediction ❤️
This work consists of the analysis of different estimators for the detection of cardiac insufficiency.
Machine Learning Models
- Logistic regression
- A Bayesian network
- KNN
- SVM with a radial kernel
- Decision tree
- Random forest
- AdaBoost
- XGBoost
- Neuronal Network
- Optimizer: Adam
- Loss: Binary Crossentropy
Transformations
- Standard scaler
- L2 normalization
Dimensionality Reduction
- PCA
- Sequential Feature Selector
- Select K Best using Chi-square
- Feature classification with recursive feature elimination (RFE)
Data visualization algorithms
- t-SNE
Data aumentations algorithms
- SMOTE
Metrics
- Accuracy
- ROC curve
- AUC
Best model
The best results were obtained with a Random Forest using the SMOTE algorithm for data augmentation and the model was trained through k-flod cross validation with k = 5
Figure 1: Results of the experiment