Practical Minimum Sample Size for Road Crash Time-Series Prediction Models

Table 3

All possible crash prediction models for Germany.

Sample size	Model	Model fit statistics
		Stationary R-squared	MAE
45	(0, 2, 1)	0.476	13208.783
	(1, 2, 0)	0.249	15495.564
	(1, 2, 1)	0.476	13143.372
	(2, 2, 0)	0.375	14227.440
	(0, 2, 2)	0.476	13136.482
	(2, 2, 2)	0.480	13020.020
	(1, 2, 2)	0.476	13159.988
	(2, 2, 1)	0.477	13143.070
	(0, 2, 3)	0.475	13089.132
	(3, 2, 0)	0.389	13912.752
	(1, 2, 3)	0.479	13160.837
	(3, 2, 1)	0.481	12830.792
	(2, 2, 3)	0.519	12101.438
	(3, 2, 2)	0.482	12830.982
35	(0, 2, 1)	0.490	12691.374
	(1, 2, 0)	0.309	14341.679
	(1, 2, 1)	0.496	12658.162
	(2, 2, 0)	0.453	12935.550
	(0, 2, 2)	0.495	12614.442
	(2, 2, 2)	0.496	12689.400
	(1, 2, 2)	0.493	12647.033
	(2, 2, 1)	0.496	12668.552
	(0, 2, 3)	0.497	12526.896
	(3, 2, 0)	0.468	12926.740
	(1, 2, 3)	0.496	12701.953
	(3, 2, 1)	0.516	12127.803
	(2, 2, 3)	0.538	12001.359
	(3, 2, 2)	0.516	12098.514
25	(0, 2, 1)	0.490	12691.374
	(1, 2, 0)	0.226	12442.211
	(1, 2, 1)	0.257	12424.256
	(2, 2, 0)	0.245	12544.195
	(0, 2, 2)	0.256	12420.705
	(2, 2, 2)	0.276	11818.040
	(1, 2, 2)	0.296	11666.041
	(2, 2, 1)	0.263	12649.132
	(0, 2, 3)	0.256	12423.869
	(3, 2, 0)	0.259	12368.022
	(1, 2, 3)	0.292	11609.888
	(3, 2, 1)	0.308	11446.555
	(2, 2, 3)	0.292	11848.622
	(3, 2, 2)	0.297	11773.639
15	(0, 2, 1)	0.595	8823.144
	(1, 2, 0)	0.429	9775.784
	(1, 2, 1)	0.661	8052.656
	(2, 2, 0)	0.495	9523.153
	(0, 2, 2)	0.702	7022.490
	(2, 2, 2)	0.665	8067.804
	(1, 2, 2)	0.649	8227.577
	(2, 2, 1)	0.672	7703.740
	(0, 2, 3)	0.715	7109.722
	(3, 2, 0)	0.625	8523.649
	(1, 2, 3)	0.702	7431.874
	(3, 2, 1)	0.709	7159.754
	(2, 2, 3)	0.728	6992.708
	(3, 2, 2)	0.712	6934.599