| Statistical Model |
Capable of modeling data, inferring relationships between variables |
Requires certain a priori assumptions, needs validation of model reliability, requires the selection of fitting data processing methods |
Hunt & Willett (2018), Tao & Michailidis (2019), Yu, Jibin & Jiang (2016)
|
| Distance model |
Can mine data in-depth |
High requirements for data preprocessing, demanding distance measurement methods, sensitive to noise |
Zhu et al. (2020), Ma, Aminian & Kirby (2019), Miao et al. (2018)
|
| Clustering model |
Broad applicability, robust interpretability |
Not suitable for high-dimensional or large-scale streaming data, sensitive to initial values, high requirements for preprocessing |
Lee & Lee (2022), Raut et al. (2023)
|
| Density model |
Simple to implement, quickly reveals potential structures and robust to noise |
Suffers from the curse of dimensionality in high-dimensional data, computationally intensive for large-scale data |
Liu et al. (2020), Zhang, Zhao & Li (2019)
|
| Isolation model |
Capable of modeling data distribution, suitable for complex data distributions |
Performance may decrease with high-dimensional data |
Liu, Ting & Zhou (2008)
|
| Frequent item mining |
Effective at identifying outliers and anomalies in low-density areas, no need for labeled data, supports unsupervised learning |
Potential for false positives due to noise and outliers in dataset |
Cai et al. (2020a), Hao et al. (2019), Cai et al. (2020b)
|
