Developer API

Interpreting Results

ZeroTrain returns a structured response that includes the decision outcome, execution details, and system feedback.

Response Structure

{
  "Header": { ... },
  "Results": [ ... ],
  "UsageInfo": { ... },
  "BatchInfo": { ... },
  "Errors": [ ... ],
  "Warnings": [ ... ]
}

Header

The Header provides high-level information about the execution.

  • Processing time and performance metrics
  • Number of rows evaluated
  • Result, error, and warning counts

Results

The Results section contains the decision outcomes. Each entry represents a model evaluation result.

{
  "modelName": "CallCenterRoutingModel",
  "decisionId": "CC-002",
  "action": "RouteToTier2",
  "confidenceScore": 0.5134,
  "logicPassed": false
}

Key Fields

  • action – The selected outcome
  • confidenceScore – Strength of alignment
  • logicPassed – Indicates whether explicit conditions were satisfied

Compact Trace Information

Trace data provides visibility into how the decision was evaluated.

"logicTrace": ["R1","C1:P","C2:F","C6:F"]
  • R1 – Evaluated row
  • Cx:P – Condition passed
  • Cx:F – Condition failed

Detailed Trace Information

The trace output may include additional context to improve readability. This format retains the compact structure while providing descriptive details for each evaluation step.

[
  "R1",
  "C1:P: CallIntent  BillingDispute",
  "C2:F: SentimentScore (null) ",
  "C4:P: HasPaymentFailure  True"
]

Format

  • R# – Evaluated row
  • C# – Column or condition identifier
  • P – Passed
  • F – Failed

Additional text may be included to describe the evaluated value or condition.

Matched Row

The matchedRow section shows values from the selected outcome.

{
  "HasPaymentFailure": "True",
  "Weight": "0.75"
}

Usage Information

UsageInfo provides execution and consumption details.

  • Rows and columns evaluated
  • Workload units consumed
  • Timing breakdown

Warnings

Warnings indicate non-blocking issues encountered during evaluation.

{
  "code": 20084,
  "message": "Null value encountered..."
}

Consistency

For the same input and model, the system will always return the same result.

Summary

ZeroTrain results provide a complete view of the decision, including outcome, traceability, and execution details.