Process Abuse recognition


The API route tries to flag users who display suspicious behaviour when using OKRA's products. It takes into account a user's profile details and record history. The API response is explained below:

  • confidence: how confident is the model
  • anomaly_score: the computed anomaly score. The lower, the more abnormal.
    *prediction: the prediction label. Values are either ABUSE or NORMAL.
  • sample_score: the computed normality score. The lower, the more abnormal.
  • explain: explanation of factors affecting the anomaly prediction
Click Try It! to start a request and see the response here!