Table 3

Accuracy of the predictive models for out-of-hospital cardiac arrest based on meteorological data, chronological data and combined meteorological and chronological data

Measure of predictive model performanceML model with comprehensive meteorological variablesML model with chronological variablesML model with combined meteorological and chronological variables
Training datasetTesting datasetTraining datasetTesting datasetTraining datasetTesting dataset
MAE by prefecture and day1.4131.6281.4151.5771.3141.547
MAPE by day (%)*12.15814.02311.30710.8337.0077.788
  • *In general, MAPE less than 10% is considered highly accurate predicting; 10%–20%: good predicting; 20%–50%: reasonable predicting; and more than 50%: inaccurate predicting.22

  • MAE, mean absolute error; MAPE, mean absolute percentage error; ML, machine learning.