An Interpretability Method of Workflow Power Consumption Based on Feature Contribution

The scheduling of workflows is one of the important tasks in cloud data centers. The power consumption of workflows is a vital factor of workflow scheduling. Workflow scheduling strategies are usually designed based on workflow power consumption. Although many high-precision models for estimating power consumption of workflows have been proposed, few of them performed well in terms of interpretability. In this paper, we propose an interpretable method-CAIM(Contribution Additive Interpretable Method) to interpret the evaluation results of the workflow power consumption models. The experimental results verify the effectiveness of our proposed method.