See What Happened

Read the audit trail, preview a run with dryRun(), and check metrics.

See What Happened

"Why was this person declined?" is the question rules engines get asked most. RuleBox can answer it three ways.

1. The audit trail

After a run, ask each rule how it went:

var book = getInstance( "LoanApproval" )
	.withDefaultResult( "MANUAL_REVIEW" )
	.run( { creditScore: 720, income: 80000, requestedAmount: 100000 } )

book.getRuleStatus( "autoApprove" )   // "EXECUTED"
book.getRuleStatusMap()
// { declineLowScores: "SKIPPED", autoApprove: "EXECUTED" }

Every rule ends a run in one of these states:

Run the same rulebook for a score of 540:

// { declineLowScores: "STOPPED", autoApprove: "REGISTERED" }

declineLowScores stopped the chain, so autoApprove was never looked at. This is why naming your rules matters.

2. Preview with dryRun()

dryRun() asks "what would happen?" without doing it. No actions run, and nothing is changed or recorded.

getInstance( "LoanApproval" )
	.dryRun( { creditScore: 720, income: 80000, requestedAmount: 100000 } )

You get one entry for each rule it reaches, in order:

[
	{ name: "declineLowScores", description: "", wouldExecute: false, wouldStop: false },
	{ name: "autoApprove",      description: "", wouldExecute: true,  wouldStop: false }
]

For a score of 540 the list has only one entry, because declineLowScores would stop the chain:

[
	{ name: "declineLowScores", description: "", wouldExecute: true, wouldStop: true }
]

Use dryRun() whenever someone asks "what happens if...?". It is safe to call on live rules.

3. Metrics

A RuleBook also keeps running totals for each rule:

book.getRuleMetrics( "autoApprove" )
// {
//   name: "autoApprove",
//   totalEvaluations: 1,
//   countsByState: { EXECUTED: 1 },
//   ... plus min, max and total duration, and first and last run times
// }

Metrics live on that one RuleBook instance. Because you ask for a fresh one each time, they are most useful for a long-lived instance. In lesson 9 the Visualizer collects them for you across every run.

Try it

Call dryRun() with a score of 600. Both rules report wouldExecute: false. That is the "manual review" case: nobody matched, so the default decides.

Next: move the rules out of code and into a file.