Welcome
User Guide
Trusted AI
This feature is available for classification and regression problem types.
Fairness Metrics
Fairness metrics are a set of measures that enables the user to detect the presence of bias in model.

Release Notes
Installation Guide
Developer Guide
MLAC User Guide
User Guide
- Introduction
  Marketing AION (Artificial Intelligence ON) is an AI life cycle management platform for applying machine learning to real-world problems. Marketing AION encompasses the complete pipeline from raw data ingestion to a deployable machine learning model with a less-code/no-code approach.
- Model Management
- Model Training
- Model Prediction
  Model prediction predicts the input data based on the pattern of training data.
- Trusted AI
  This feature is available for classification and regression problem types.
  - Fairness Metrics
    Fairness metrics are a set of measures that enables the user to detect the presence of bias in model.
    - Viewing Fairness Metrics
      To view the fairness metrics, complete the following steps:
  - Performance
    Model performance is an assessment of the model's ability to perform a task accurately not only with training data but also in real-time with runtime data when the model is deployed.
  - Transparency
    Transparency helps users to understand the input data, algorithm and expected output of an ML model.
  - Explainability
    Explainability tab gives an explanation about Prediction and Model.
  - Robustness
    Robustness of AI/ML models is analyzed by Sensitivity Analysis examining the extent to which the results are affected by changes in input data.
- Model Monitoring
  This feature is available for classification and regression problem types.
- Utility
  For Data Generator and BYOM (Bring Your Own Model), select the Utility tab.

Fairness Metrics

Fairness metrics are a set of measures that enables the user to detect the presence of bias in model.


Theil Index	Measures inequality among individuals in the group for a classification problem type. Value of Theil Index is always greater than zero. Ideally it should be zero for perfect fairness and biasness increase with increase in value higher than zero.
Equal Opportunity Difference	Measures the difference in true positive rate between privileged and unprivileged group of protected features for a classification problem type. Ideally it should be zero to be fair. A value less than zero illustrates a higher benefit for the privileged group and positive value is a sign of higher benefit for the unprivileged group. Statistical Parity Difference measure the difference in the rate of favorable outcomes between privileged and unprivileged groups. This measure must be equal to 0 to be fair.