Skip to main content
. 2015 Sep 8;15(9):22616–22645. doi: 10.3390/s150922616

Table 4.

Prediction accuracy improvement, by applying temporal granularity (TG) on timestamp and semantic abstraction (SA) on sensor data.

Classification Algorithm BaseLine TG TG-SA
Naive Bayes ROC-area 0.82 0.93 0.94
Precision 0.56 0.65 0.66
Recall 0.69 0.73 0.74
F-Measure 0.58 0.65 0.68
SVM ROC-area 0.79 0.86 0.86
Precision 0.70 0.69 0.72
Recall 0.74 0.75 0.78
F-Measure 0.67 0.73 0.75
Decision Tree ROC-area 0.72 0.70 0.72
Precision 0.46 0.46 0.55
Recall 0.65 0.64 0.68
F-Measure 0.54 0.54 0.57
Logistic
Regression
ROC-area 0.79 0.83 0.89
Precision 0.66 0.69 0.64
Recall 0.71 0.75 0.78
F-Measure 0.64 0.65 0.68