2015 年 41 巻 p. 80-87
In recent years, it is remarkable that the increase in power consumption of the consumers belonging to business sector. The consumers belonging the business sector has been introduced demand rate system, so the reduction of power costs by reducing the maximum demand value is noted. Therefore, understanding of the maximum demand value by the prediction of the demand value is important. The conventional prediction, there is method using a time series model, it is effective when there is a periodicity in the data. However, since the power consumption is greatly affected by external factors, using the time series model is difficult to predict. In this paper, we do prediction using a decision tree analysis can be considered the external factors. we focus on the correlation between the demand value data, and the temperature that has a high relationship to power consumption. We performed the power consumption prediction by applying the decision tree and evaluated the accuracy. As a result, it is possible to verify the change in prediction accuracy due to changes in the analytical conditions, and we investigated for improving the accuracy.