Activity-based Twitter sampling for content-based and user-centric prediction models

Table 3 The prediction performance for content-based over 7 lags

	Lag \(=\) 1	Lag \(=\) 2	Lag \(=\) 3	Lag \(=\) 4	Lag \(=\) 5	Lag \(=\) 6	Lag \(=\) 7
Activity-based
Narcotics	0.51	0.54	0.52	0.53	0.58	0.53	0.67
Deceptive	0.65	0.52	0.57	0.64	0.65	0.62	0.51
Criminal damage	0.43	0.6	0.7	0.65	0.6	0.56	0.54
Burglary	0.52	0.58	0.56	0.56	0.56	0.54	0.52
Battery	0.61	0.7	0.62	0.72	0.67	0.66	0.6
Assault	0.46	0.47	0.57	0.52	0.5	0.54	0.56
Prostitution	0.57	0.59	0.7	0.68	0.68	0.58	0.68
PublicViolation	0.46	0.51	0.47	0.51	0.55	0.55	0.53
Robbery	0.55	0.56	0.47	0.55	0.52	0.52	0.56
Theft	0.65	0.55	0.52	0.6	0.62	0.58	0.62
All	0.77	0.74	0.7	0.86	0.76	0.7	0.73
Random
Narcotics	0.5	0.51	0.57	0.55	0.55	0.55	0.65
Deceptive	0.63	0.56	0.55	0.55	0.68	0.67	0.60
Criminal damage	0.42	0.62	0.69	0.68	0.65	0.56	0.51
Burglary	0.53	0.5	0.52	0.53	0.57	0.54	0.52
Battery	0.46	0.67	0.65	0.71	0.7	0.7	0.57
Assault	0.46	0.54	0.56	0.54	0.53	0.52	0.55
Prostitution	0.62	0.62	0.67	0.67	0.68	0.64	0.54
PublicViolation	0.44	0.47	0.46	0.46	0.47	0.53	0.49
Robbery	0.5	0.54	0.51	0.56	0.5	0.49	0.44
Theft	0.57	0.57	0.56	0.53	0.61	0.58	0.55
All	0.5	0.59	0.6	0.59	0.54	0.58	0.61

The italic emphasizes show in which experiments the activity-based or random sampling performed better