Skip to main content

Table 2 Forecasting results

From: Activism via attention: interpretable spatiotemporal learning to forecast protest activities

  Charlottesville Ferguson I Ferguson II
F-score AUC F-score AUC F-score AUC
LR[tem] 0.200 0.696 0.103 0.733 0.343 0.752
LR[s,tem] 0.182 0.789 0.259 0.766 0.327 0.789
LR[s,tem,st] 0.200 0.734 0.230 0.722 0.314 0.773
SVM[tem] 0.200 0.818 0.000 0.791 0.400 0.816
SVM[s,tem] 0.186 0.809 0.000 0.796 0.408 0.837
SVM[s,tem,st] 0.000 0.782 0.000 0.754 0.313 0.780
LSTM 0.240 0.752 0.415 0.801 0.417 0.819
S + LSTM 0.267 0.778 0.423 0.804 0.439 0.838
S + LSTM (GL) 0.308 0.793 0.423 0.805 0.440 0.839
RMTFL 0.182 0.663 0.250 0.703 0.250 0.829
CMTFL-1 0.182 0.664 0.350 0.711 0.316 0.805
CMTFL-2 0.200 0.661 0.333 0.711 0.324 0.815
ActAttn (w/o GL) 0.308 0.830 0.459 0.820 0.464 0.849
ActAttn (w/o stAttn) 0.324 0.797 0.406 0.783 0.409 0.842
ActAttn (w/o spAttn) 0.333 0.836 0.448 0.812 0.448 0.846
ActAttn 0.400 0.843 0.462 0.822 0.471 0.853