From: Healthcare predictive analytics using machine learning and deep learning techniques: a survey
Method | Advantages | Disadvantages |
---|---|---|
Linear regression | • Linear regression models are easy to understand for beginners • Training linear regression models is fast, even on large datasets • Linear regression models can forecast, classify, and predict • Eliminating overfitting by regularization | • Linearity: Linear regression models require linearity between independent and dependent variables. This can limit nonlinear relationships •  It is not recommended for most practical applications as it greatly simplifies real- world problems |
Logistic regression | • Excellent performance with small datasets • Its output is interpretable as probability | • Compliant data assumptions are required • It only offers linear solutions |
Decision trees | • They can manage categorical characteristics • There are a few parameters to tune • They perform well with large feature-count datasets | • The interpretability of the ensemble is questionable |
Random forest | • Even with noisy or imbalanced data, random forest can achieve high accuracy • Robustness to overfitting: random forest generalizes well to new data • Interpretability: Random forest models are easy-to-understand • Random forest scales to large datasets | • Computational complexity: Random forest training is computationally expensive, especially for large datasets • Sensitivity to hyperparameters: random forest performance can be sensitive to hyperparameters |
Support vector machine | • High-dimensional space for input • Few irrelevant features • Document vectors are sparse | •Data collection is time-consuming |
K-nearest neighbors | • Simple algorithm | • The user must specify the number of neighbors • A high level of relative computational complexity |
Naive Bayes | • Simple and straightforward method • It combines effectiveness and reasonable precision | • Used primarily when the size of the training set is smaller • It assumes the conditional independence of linguistic features |