Abstract: A non-transitory computer-readable storage medium storing a prediction program for causing a computer to perform processing including: listing combinations of feature amounts that are correlated with a target label; creating a policy to achieve the target label for a prediction target based on a difference between the listed combinations of the feature amounts and a combination of feature amounts of the prediction target; and determining appropriateness of the created policy based on performance information that indicates past performances.

Abstract: A model-match-rate evaluating unit of a transaction monitoring device, which monitors a transaction system, evaluates a ratio of the number of transactions that match any models and respective processing times of all layers in the transaction are each within a corresponding normal range to the number of transactions observed per unit time as a model match rate. When the model-match-rate evaluating unit detects an abnormality of the system based on the model match rate, a suspicious-point-in-suspicious-model extracting unit of a transaction detail analyzing device extracts a point where a processing time deviates from the normal range as a suspicious point, a problematical-point evaluating unit evaluates a problem of each suspicious point as a problematical point, and a detail-analysis-result display unit displays an evaluation result of the problematical point and the suspicious point.

Abstract: A system analyzing apparatus analyzes a processing state of a computer system that includes a plurality of types of unit processing. The system analyzing apparatus includes a selecting unit that selects a plurality of representative unit processing based on similarities between individual units of the unit processing; and a grouping unit that groups each unit processing based on similarity between the each unit processing and each of representative unit processing selected by the selecting unit.

Abstract: In determining a condition of a case under which the occurrence rate of the performance abnormality is high based on information related to a resource utilization situation acquired in a resource utilization situation acquisition step and an occurrence frequency of a request acquired in a request occurrence frequency acquisition step, and information related to the existence/nonexistence of the performance abnormality determined in a performance abnormality existence/nonexistence determination step, when branching a parent node that represents a situation of a decision tree to two child nodes using the method of decision tree, the decision tree is generated using an evaluation function which performs branching such that most of normal samples which belong to the parent node are contained in the one child node, and most of abnormal values which belong to the parent node are contained in the other child node.

Abstract: A model-match-rate evaluating unit of a transaction monitoring device, which monitors a transaction system, evaluates a ratio of the number of transactions that match any models and respective processing times of all layers in the transaction are each within a corresponding normal range to the number of transactions observed per unit time as a model match rate. When the model-match-rate evaluating unit detects an abnormality of the system based on the model match rate, a suspicious-point-in-suspicious-model extracting unit of a transaction detail analyzing device extracts a point where a processing time deviates from the normal range as a suspicious point, a problematical-point evaluating unit evaluates a problem of each suspicious point as a problematical point, and a detail-analysis-result display unit displays an evaluation result of the problematical point and the suspicious point.

Abstract: A service process analyzing program, stored in a computer-readable medium, for identifying service transactions taking place between servers, particularly capable of discriminating between different transaction classes, including infrequent classes. A definite transaction finder identifies definite transactions by extracting a combination of messages having unique and definite call dependency between them. A transaction candidate builder extracts a plurality of combinations of messages including an unsettled message, thereby producing one or more transaction candidates. A candidate evaluator evaluates the likelihood of each transaction candidate, based on how many transactions of the same class have been identified as definite transactions or estimated transactions. A transaction estimator produces another estimated transaction by selecting one of the transaction candidates based on their respective likelihood scores.

Abstract: In determining a condition of a case under which the occurrence rate of the performance abnormality is high based on information related to a resource utilization situation acquired in a resource utilization situation acquisition step and an occurrence frequency of a request acquired in a request occurrence frequency acquisition step, and information related to the existence/nonexistence of the performance abnormality determined in a performance abnormality existence/nonexistence determination step, when branching a parent node that represents a situation of a decision tree to two child nodes using the method of decision tree, the decision tree is generated using an evaluation function which performs branching such that most of normal samples which belong to the parent node are contained in the one child node, and most of abnormal values which belong to the parent node are contained in the other child node.

Abstract: A system monitoring apparatus, method and program determining whether each unit process executed by a system matches one of a plurality of unit process models generated by individually modeling a plurality of types of unit processes, determining whether an analysis of a processing status of the system is needed based on determining whether each unit process matches one of the plurality of unit process models, and displaying a difference of a most approximate unit process model among the plurality of unit process models in relation to each unit process determined as matching none of the plurality of unit process models.

Abstract: A system analyzing apparatus analyzes a processing state of a computer system that includes a plurality of types of unit processing. The system analyzing apparatus includes a selecting unit that selects a plurality of representative unit processing based on similarities between individual units of the unit processing; and a grouping unit that groups each unit processing based on similarity between the each unit processing and each of representative unit processing selected by the selecting unit.

Abstract: A service process analyzing program, stored in a computer-readable medium, for identifying service transactions taking place between servers, particularly capable of discriminating between different transaction classes, including infrequent classes. A definite transaction finder identifies definite transactions by extracting a combination of messages having unique and definite call dependency between them. A transaction candidate builder extracts a plurality of combinations of messages including an unsettled message, thereby producing one or more transaction candidates. A candidate evaluator evaluates the likelihood of each transaction candidate, based on how many transactions of the same class have been identified as definite transactions or estimated transactions. A transaction estimator produces another estimated transaction by selecting one of the transaction candidates based on their respective likelihood scores.