Intelligent Behavior Data Analysis for Internet Addiction

Table 4

Regression task.

Feature model	ia−	ia (LIA)	ia (NIA)	ia (CIA)
DT	0.000076	0.000061 (19.7%)	0.000072 (5.3%)	0.000064 (15.8%)
SVM	0.004114	0.003636 (11.6%)	0.003677 (10.6%)	0.003024 (26.5%)
KNN	0.000065	0.000064 (1.5%)	0.000066 (−1.5%)	0.000064 (1.5%)
RF	0.000040	0.000039 (2.5%)	0.000040 (0%)	0.000038 (5%)
GBDT	0.000042	0.000039 (7.1%)	0.000042 (0%)	0.000040 (4.8%)
Bagging	0.000041	0.000039 (7.3%)	0.000041 (0%)	0.000039 (7.3%)
ET	0.000068	0.000057 (16.2%)	0.000065 (4.4%)	0.000065 (4.4%)

“ia−” refers to the baseline experiment and “ia (LIA)” stands for the experiment with the internet addiction value learned by the naive LIA model, which gets the best results in the internet addiction calculation task using the LIA model. “ia (NIA)” represents the experiment with the best internet addiction value learned by the NIA model without regular behavior consideration, which gets the best results in the internet calculation task using the NIA model. Similarly, “ia (CIA)” refers to the experiment with the internet addiction value learned from the clustering-based model.