| Traditional models | HMLS | Machine learning algorithms |
---|---|---|---|
Strengths | Simple, easy to understand and master | Improve extrapolation and interpretation of models | 1. Applicable to datasets with diverse distribution 2. Robust to outliers and missing values 3. Allow for direct feature selection 4. Avoid over-fitting 5.Strong extrapolation capabilities |
Weaknesses | 1. The sensitivity to outliers 2. A weak feature selection ability 3. Overfitting 4. Limited ability to extrapolate beyond the available data | 1. Filter out irrelevant features depends on prior knowledge and expertise. 2. Weak ability to handle non-linear relationships 3.May not fit accurately for complex data patterns. | Filter out irrelevant features depends on prior knowledge and expertise |