For extrapolation of injury predictions far into the future (e.g. 2040) we will need to consider how we want to the time trend to taper.
For reference, here are the coefficients for the linear 'year' covariate from the most recent injury models.
(Define modelled modes = car, bike, motorcycle, walk. Other modes = NOV, truck, bus, van.)
(1) Modelled vs modelled: -0.020
(2) Modelled cas vs other strike: -0.034
(3) Modelled strike vs other cas: -0.043
(4) Other cas vs other strike: -0.038
To project X years into the future, multiply the current value by exp ( coef * X ). E.g. if we project all models 20 years into the future we expect the number of injuries to decrease to
For extrapolation of injury predictions far into the future (e.g. 2040) we will need to consider how we want to the time trend to taper.
For reference, here are the coefficients for the linear 'year' covariate from the most recent injury models.
(Define modelled modes = car, bike, motorcycle, walk. Other modes = NOV, truck, bus, van.)
(1) Modelled vs modelled: -0.020
(2) Modelled cas vs other strike: -0.034
(3) Modelled strike vs other cas: -0.043
(4) Other cas vs other strike: -0.038
To project X years into the future, multiply the current value by exp ( coef * X ). E.g. if we project all models 20 years into the future we expect the number of injuries to decrease to
(1) 67%
(2) 51%
(3) 42%
(4) 47%