Data mining – DM

This table function returns the sequence rules that are contained in a sequence rule model. Each row in the table represents one sequence rule.

Syntax

Function syntax


>>-DM_getSeqRules--(--ruleModel--)-----------------------------><

Parameters

ruleModel: A value of type DM_RuleModel

Return Value

The return value is of type TABLE

If ruleModel is not of type SequenceModel, the following error occurs:
IDM1101E: For the method DM_getSeqRules, the input model is not valid. Valid models for this method are PMML models of the type SequenceModel.
If the model is inconsistent, for example, the item set information in the model contains an invalid item, or the sequence information refers to an invalid item set, or the sequence rule information refers to an invalid sequence. , the following error occurs:
IDM1103E: While parsing an association rule model or a sequence rule model for the method DM_getSeqRules, an error occurred. The model is not a rule model, or it is damaged.

Table 1. Result Values for DM_getSeqRules
Column	Data type	Purpose
ruleId	INTEGER	The unique identifier of a sequence rule. An integer ID that runs from 1 to the number of sequence rules in the model.
bodySeqId	INTEGER	The unique identifier of the sequence that forms the body or the antecedent part of a sequence rule. An integer ID that runs from 1 to the number of sequences in the model.
headSetId	INTEGER	The unique identifier of the sequence that forms the head or the consequent part of a sequence rule. An integer ID that runs from 1 to the number of sequences in the model.
bodySeqText	VARCHAR(4096)	The body or the antecedent part of a sequence rule in textual form. The items in this text are represented by their mapped names. Items that form a single item set are concatenated by '+'. Adjacent item sets of the sequence are concatenated by the symbol '>>>'. This symbol indicates a time step.
headSetText	VARCHAR(4096)	The head or consequent part of a sequence rule in textual form. The items in this text are represented by their mapped names. Items that form a single item set are concatenated by '+'. Adjacent item sets of the sequence are concatenated by the symbol '>>>'. This symbol indicates a time step.
length	INTEGER	The number of item sets in a sequence rule including the item set in the rule head.
nbOfItems	INTEGER	The total number of items in a sequence rule including the items in the rule head.
support	DOUBLE	The support of a sequence rule. The support value is a number between 0 and 1. It represents the ratio between the number of transaction groups that support (contain) the sequence rule and the total number of transaction groups.
confidence	DOUBLE	The confidence of a sequence rule. The confidence value is a number between 0 and 1. It represents the ratio between the number of transaction groups that support (contain) the entire sequence rule and the number of transaction groups that support (contain) the body of a sequence rule.
lift	DOUBLE	The lift of a sequence rule. The lift value is a positive number that indicates the ratio between the actual support of a rule and its 'expected' or predicted support. It can be calculated based on the assumption that the distribution of the item sets in the input data is completely randomly and statistically independent from other item sets.
meanTimeDiff	DOUBLE	The mean elapsed time between the beginning and the end of a sequence rule. This quantity is calculated from the training data in the following steps: Every transaction group that supports the rule is traversed from the lowest to the highest group value (time stamp). The first combination of item sets that supports the entire rule and that matches the filter constraints of the mining task is detected. The maximum difference of the group values (time stamp) in the detected list of item sets is calculated. The average of these maximum group-value differences is returned as 'meanTimeDiff'.
stdDevTimeDiff	DOUBLE	The standard deviation of the elapsed time distribution between the beginning and the end of a sequence rule. This quantity is calculated from the training data in the following steps: Every transaction group that supports the rule is traversed from the lowest to the highest group value (time stamp). The first combination of item sets that supports the entire rule and that matches the filter constraints of the mining task is detected. The maximum difference of the group values (time stamp) in the detected list of item sets is calculated. The standard deviation of these maximum group-value differences is returned as 'stdDevTimeDiff'.