4.1 Fields

This topic provides information on Fields.

This section allows users to define basic model details.

Description of the illustration anolmodel8.png

• Use Case Name:
  • Enter a unique name for the model.
  • Example: "Login_Anomaly_Model" or "Payment_Fraud_Detection"
  • (Required) – This field must be filled to proceed.
• Description:
  • Provide a summary of the model’s purpose.
  • Example: "Detects unusual login attempts based on user behaviour patterns."
• Use Case Type:
  • Select the type of use case as Anomaly_Detection.
  • Options may Regression & Classification, or any other specific use cases.(Required)
• Product Processor:
  • Select the system or processor that will handle training.
  • Example: "OBDX"
  • (Required)
• Training Data Source:
  • Specify the dataset used to train the anomaly detection model.
  • The dataset must include the target column (i.e., the column indicating whether an instance is anomalous or normal).
  • Example: A CSV file or database table containing past login records.
  • (Required)
• Inference Data Source:
  • Specify the dataset used when making predictions.
  • Unlike the training dataset, this dataset should not include the target column.
  • Example: "Live payment transaction records without labels."
  • (Required)
• Unique Case Identifier:
  • Select the column in the dataset that uniquely identifies each record.
  • Example: "User_ID" for login data or "Transaction_ID" for payment data.
  • (Required)
• Target Column:
  • Select the column that defines whether a transaction/login attempt is an anomaly.
  • Example: A column labelled "Anomaly_Flag" where 1 indicates an anomaly and 0 indicates normal behaviour.
  • (Required)
• Positive Target Value:
  • Specify the value that represents an anomaly.
  • Example: If "1" indicates fraud or an unauthorized login, set "1" as the positive target value.
• Tablespace:
  1. Define the storage location for the model’s data within the system.
• Partition Column Names:
  • Select the columns used for partitioning the dataset.
  • Example: "Date" to separate records by time period.
• Selected Algorithm:
  • Choose the machine learning algorithm to be used.
  • Example: ALGO_SUPPORT_VECTOR_MACHINES, ALGO_NEURAL_NETWORK etc.
• Model Error Statistic:
  • Select an error metric to evaluate the model’s accuracy.
  • Example: F1 Score, Precision-Recall, or AUC-ROC.
• Correlation Button:
  • Clicking this button will analyse relationships between features and the target variable.
  • Helps in understanding the significance of different input features.
• Cost Matrix Button:
  • Allows users to define cost-sensitive learning, useful for reducing false positives or false negatives.
  • (Optional)
• Save Button:
  1. Saves the model configuration.
• Cancel Button:
  • Exits without saving any changes.