Call generation config file

The simulation can either use a calls csv file (e.g. calls.csv) by specifying <calls>calls.csv</calls> under <simConfig><simFiles> in a sim config (e.g. sim_config.xml), or the calls can be generated based on parameters in a call generation config file using <callGenConfig>gen_config.xml</callGenConfig>. For an example of such a file, see JEMSS/data/cities/auckland/models/1/calls/reps/train/gen_config.xml.

Below explains the contents of the call generation config file:

<genConfig>
	<mode>calls</mode>

	<!-- number of call sets to be generated, here we use 30 replications -->
	<numCallsFiles>30</numCallsFiles>

	<!-- where calls files will be written to file (if they are written, which is controlled by sim_config.xml) -->
	<outputPath>output</outputPath>
	
	<sim>
		<!-- time (in days) after which calls will start being generated, usually leave this as 0.0 -->
		<startTime>0.0</startTime>

		<!-- earliest time (in days) that last call will arrive -->
		<minLastCallArrivalTime>1 + 14 + 0.1</minLastCallArrivalTime>
		
		<callDistributions>
			<!-- distribution for duration between call arrivals, below is for 280 calls per day -->
			<interarrivalTime seed="1217248832">Exponential(1/280)</interarrivalTime>

			<!-- priority/urgency of calls, below is for 57% high, 24.5% medium, and 18.5% low priority -->
			<priority seed="199894915">Categorical([0.57, 0.245, 0.185])</priority>

			<!-- delay (in days) between call arrival and ambulance dispatch,
			below is for normal distribution with 50 seconds mean and standard deviation 0 -->
			<dispatchDelay seed="293110701">Normal(50/60/60/24, 0)</dispatchDelay>

			<!-- on-scene or treatment duration (in days), below is for 12 minutes average -->
			<onSceneDuration seed="2023374599">Exponential(12/60/24)</onSceneDuration>

			<!-- probability that transport to hospital is required, below is for 80% probability -->
			<transport seed="1427137433">Bernoulli(0.8)</transport>

			<!-- duration (in days) that ambulance spends upon reaching hospital for patient handover -->
			<handoverDuration seed="1330644641">Exponential(12/60/24)</handoverDuration>
		</callDistributions>
		
		<!-- call spatial density distribution, should be a raster with WGS84 coordinates -->
		<callDensityRaster cellSeed="1442623348" cellLocSeed="1963951879">
			<filename>$jemssDir/data/cities/auckland/data/population density/census 2013/rasters/urban/1/auckland-urban-pop-densit-40m.tif</filename>
		</callDensityRaster>
	</sim>
</genConfig>

The statistical distributions used above like Exponential and Normal are provided by Distributions.jl, through which many other distributions are available.

The seed attributes (seed, cellSeed, cellLocSeed) are for seeding the random number generators so that the generated calls are reproducible. These seed attributes are optional but it is recommended to specify seed values (rather than omitting them), but make sure that different seeds are used for training, validation, and testing data.