Introduction

The Income Estimator predicts the likelihood that a user belongs to a higher income segment based on behavioral data.

Unlike traditional income verification, this model does not predict an individual's exact monthly income. Instead, it estimates the probability that a user belongs to a higher income group and maps that prediction into predefined income tiers.

The API returns:

• probability — the model's prediction score (recommended for decisioning)
• tier — simplified income category
• confidence — model certainty
• incomeRange — reference range associated with the predicted tier

---

Understanding the Outputs

1. Probability (Primary Output)

   The probability field is the primary output of the model and contains the most predictive information.
   It represents the likelihood that the user belongs to the High-income segment.

   Example:

   Probability	Interpretation
   0.95	Very strong High-income signal
   0.82	Strong High-income signal
   0.64	Moderate High-income signal
   0.48	Near decision boundary
   0.18	Strong Non-High-income signal
   Two users may receive the same income tier while having different probabilities.

   Example:

   User	Probability	Tier
   User A	0.91	High
   User B	0.63	High
   Although both users are classified as High, User A has a substantially stronger model signal.

   • Recommendation: Use probability as your primary decisioning signal.

2. Tier

   The tier provides a simplified categorization of the model output.

   Current tiers include:

   Tier	Monthly Income Reference
   Low	≤ 17,250
   Middle	17,251 – 29,250
   High	≥ 29,250
   The tier is intended for:

   • Customer segmentation
   • User interface display
   • Reporting
   • Simple business rules
     The tier should not be interpreted as an exact estimate of a customer's monthly income.

3. Income Range

   The incomeRange object represents the predefined range associated with the predicted tier.

   For example:

   Text

   {
       "monthlyIncomeMin":17251,
       "monthlyIncomeMax":29250
   }

   This range is not personalized.

   Every user classified into the same tier receives the same range.

   The values represent the tier boundaries rather than an estimated income for the individual customer.

4. Confidence

   The confidence field measures how far the prediction is from the model's internal decision boundary.
   It does not indicate the likelihood that a customer belongs to a particular income tier.

   For example:

   Probability	Confidence
   0.90	High
   0.52	Low
   0.10	High
   This may appear counterintuitive because confidence measures model certainty, not income likelihood.

   For example:

   • Probability = 0.10 means the model believes the customer is unlikely to be a High-income user.
   • Confidence = High means the model is very certain about that prediction.

---

Recommended Usage

Primary Decision Signal

Use probability for automated decisioning.

Example:

Thresholds should be calibrated based on your organization's risk appetite and historical performance.

Secondary Decision Signal

Use tier for:

• Customer segmentation
• Portfolio reporting

Avoid

The Income Estimator should not be used as:

• An exact income prediction
• A substitute for verified income documentation
• The sole approval criterion for lending decisions