Increase usefulness of model trace facilities

[mabcal]

This includes the following:

1. The current tracing of model outcomes produces too much output. Make it possible to limit it in two ways. (a) Make it subject to the existing configuration parameter TraceAfterInvalidAttempts, so it will only report the outcomes for person-days or household-days where there have been a certain number of invalid attempts. (b) Have a new user input for a TraceHousehold. If this is given a value greater than 0, it will only produce the model trace output for that particular household.

2. In addition to being able to write out model choice outcomes, add write statements for key model calculations from ChoiceProbabilityCalculator. This includes each AddUtilityTerm, AddUtilityComponent, and AddAlternative operation, and then in SimulateChoice, the drawing of a random number and the choice probabilities of all the alternatives. This can have it's own configuration control to turn it on and off, subject to the same conditions as writing the model outcomes (specific household or # of invalid attempts). This restores the trace functionality that was once in DaySim and was then removed.

3. This will work from either Default or Actum.

[PeterVermont]

Some of the ChoiceProbabilityCalculator debugging you mention are in place using a local choiceprobabilty field called intenseDebugging. Be warned though - this generates a lot of data.

…

________________________________ From: Mark Bradley <[email protected]> Sent: Friday, October 26, 2018 6:58:06 PM To: RSGInc/DaySim Cc: Subscribed Subject: [RSGInc/DaySim] Increase usefulness of model trace facilities (#181) This includes the following: 1. The current tracing of model outcomes produces too much output. Make it possible to limit it in two ways. (a) Make it subject to the existing configuration parameter TraceAfterInvalidAttempts, so it will only report the outcomes for person-days or household-days where there have been a certain number of invalid attempts. (b) Have a new user input for a TraceHousehold. If this is given a value greater than 0, it will only produce the model trace output for that particular household. 2. In addition to being able to write out model choice outcomes, add write statements for key model calculations from ChoiceProbabilityCalculator. This includes each AddUtilityTerm, AddUtilityComponent, and AddAlternative operation, and then in SimulateChoice, the drawing of a random number and the choice probabilities of all the alternatives. This can have it's own configuration control to turn it on and off, subject to the same conditions as writing the model outcomes (specific household or # of invalid attempts). This restores the trace functionality that was once in DaySim and was then removed. 3. This will work from either Default or Actum. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub<#181>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AA2RkXgGPd1M-JjrhiZ6WDCw4dlwGj9aks5uo5N-gaJpZM4X9Fov>.

[mabcal]

Thanks, I imagine I’ll use that – maybe give the user a couple different options for how much detail they want. From: Peter Andrews <[email protected]> Sent: Friday, October 26, 2018 8:24 PM To: RSGInc/DaySim <[email protected]> Cc: Mark Bradley <[email protected]>; Author <[email protected]> Subject: Re: [RSGInc/DaySim] Increase usefulness of model trace facilities (#181) Some of the ChoiceProbabilityCalculator debugging you mention are in place using a local choiceprobabilty field called intenseDebugging. Be warned though - this generates a lot of data.

…

________________________________ From: Mark Bradley <[email protected]<mailto:[email protected]>> Sent: Friday, October 26, 2018 6:58:06 PM To: RSGInc/DaySim Cc: Subscribed Subject: [RSGInc/DaySim] Increase usefulness of model trace facilities (#181) This includes the following: 1. The current tracing of model outcomes produces too much output. Make it possible to limit it in two ways. (a) Make it subject to the existing configuration parameter TraceAfterInvalidAttempts, so it will only report the outcomes for person-days or household-days where there have been a certain number of invalid attempts. (b) Have a new user input for a TraceHousehold. If this is given a value greater than 0, it will only produce the model trace output for that particular household. 2. In addition to being able to write out model choice outcomes, add write statements for key model calculations from ChoiceProbabilityCalculator. This includes each AddUtilityTerm, AddUtilityComponent, and AddAlternative operation, and then in SimulateChoice, the drawing of a random number and the choice probabilities of all the alternatives. This can have it's own configuration control to turn it on and off, subject to the same conditions as writing the model outcomes (specific household or # of invalid attempts). This restores the trace functionality that was once in DaySim and was then removed. 3. This will work from either Default or Actum. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub<#181>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AA2RkXgGPd1M-JjrhiZ6WDCw4dlwGj9aks5uo5N-gaJpZM4X9Fov>. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub<#181 (comment)>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AOwsPK3JxeLq_jig7bS29DCqZKmG8Ym0ks5uo9HZgaJpZM4X9Fov>.