MicroStrategy ONE

Model tab, Information sub-tab

The Information sub-tab contains data from the PMML description of the Model and parameters from the predictive metric. Information gathered from the PMML includes:

  • PMML Version: The version of the PMML standard used by the model

  • Application Copyright: Copyright information provided when the model was created

  • Application: The data mining tool that was used to develop the model

  • Application Version: The version of the data mining tool that was used to develop the model

  • Timestamp: Reference to the data and time the model was created (optional, format varies by vendor)

  • Model Type: The type of model, such as Neural Network, Regression, or Tree

  • Model Name: Name of the model provided by the model developer

  • Function Name: The type of function implemented by the model, such as regression or classification

  • Algorithm Name: The name of the algorithm used by the model

  • Type of split: Either Binary or Multiple, for Tree models

  • Activation Function: For Neural Network models only

  • Number of Layers: For Neural Network models only

  • Threshold: For Neural Networks and Naïve-Bayes models only

  • Regression Type: For Regression models only

  • Target Field: For Regression models only

  • Normalization Method: For Regression models only

  • Target Variable: For General Regression models only

  • Model Class: For Clustering models only

  • Number of Clusters: For Clustering models only

  • Number of Transactions: For Association and Sequence models only

  • Number of Items in Largest Transaction: For Association models only

  • Average Number of Items per Transaction: For Association models only

  • Number of Events in Largest Transaction: For Sequence models only

  • Average Number of Events per Transaction: For Sequence models only

  • Minimum Relative Support (all rules): For Association and Sequence models only

  • Minimum Confidence for (all rules): For Association and Sequence models only

  • Length Limit: For Association and Sequence models only

  • Number Of Items: For Association and Sequence models only

  • Number Of Item sets: For Association and Sequence models only

  • Number Of Sequences: For Sequence models only

  • Number Of Rules: For Association and Sequence models only

  • Time Window Width: For Sequence models only

  • Minimum Time: For Sequence models only

  • Maximum Time: For Sequence models only

  • Model Accuracy: Provides a representation of the accuracy of logistic regression models created by MicroStrategy. The value is calculated as follows: number of correct results / total number of training records.

Only a portion of these are included, depending on the model type and the information included by the tool that generated the model. For more detailed information, see the PMML documentation on the Data Mining Group website.

The Model information table also includes information from the predictive metric parameters, including:

  • PmmlVersion: The version of the PMML used by the model.

  • ModelID: A unique numeric ID assigned to this model.

  • Confidence: Used by models which do not define outputs. A value of false specifies that the model should return a score, while a value of true indicates that the confidence of the score should be returned.

  • Target: The target is significant only when Confidence has a value of true, and the predicted variable is categorical. This field can be used to specify that the model should return the confidence of a particular class of the predicted variable. If this field is left blank, the confidence of the winning class is returned.