Table 2.
ARIMA forecasting models specifications and accuracy measures
| Specification | Pre–COVID–19 | Post–COVID–19 | |||
|---|---|---|---|---|---|
| ME | MAPE | ME | MAPE | ||
| All crimes | ARIMA (3,0,0)(0,1,1) [7] | 3.75 | 6.30 | −191.27 | 59.08 |
| Violent robbery | ARIMA (1,0,1)(2,1,1) [7] | 1.42 | 9.11 | −51.25 | 60.35 |
| Non-violent robbery | ARIMA (1,1,1)(2,0,0) [7] | 1.18 | 12.19 | −26.96 | 62.04 |
| Robbery against residence | ARIMA (0,1,1)(2,0,0) [7] | 0.56 | 22.33 | −7.56 | 163.38 |
| Serious violent crime (non-sexual) | ARIMA (2,1,1)(2,0,0) [7] | 0.85 | 18.30 | −8.44 | 60.15 |
| Sexual violence | ARIMA (1,0,1) w/linear trend | 0.82 | 35.83 | −15.29 | 196.12 |
| Domestic violence | ARIMA (2,0,1)(2,0,0) [7] w/linear trend | 1.01 | 14.77 | −21.90 | 51.03 |
| VAW helpline calls | ARIMA (0,0,2)(1,0,1) [7] w/linear trend | 1.12 | 18.08 | 17.60 | 24.36 |
| BRT + SCT passengers (in millions) | ARIMA (4,0,0)(2,0,0) [7] w/mean | 0.10 | 16.39 | −2.77 | 191.91 |
ME is the mean forecast error, MAPE is the mean absolute percentage error