Associate professor Kei Hirose from Kyushu University COI Mathematics-for-Industry Group (Institute of Mathematics for Industry, Kyushu University) developed a power demand forecasting software using a mathematical model.

This software forecasts daily power demands using data of past power demands and the temperature high (or average) of the given day. It features:
- Relatively high forecasting accuracy compared to most existing machine learning methods （reference）
- High-speed computing based on regression model
- Interpretation of effects temperature has on power demands
- High versatility, such as factoring in the COVID-19 influence（reference)
- Prediction interval settings

 These features are realistic requirements.

I’m Kei Hirose from the Institute of Mathematics for Industry.

I’m very pleased to release this software. It was difficult to create everything from statistic models to GUI. But after all, it was best if someone who understood the analysis method created the software, so I didn’t give up. The user interface is made as simple as possible, so that the analysis method is easy to use.

I had numerous discussions on analysis algorithm with Prof. Hiroki Masuda from Kyushu University Faculty of Mathematics. I have received comments from Kyudai COI members in developing this tool. I couldn’t have released this software without the support and guidance from all of you. Thank you very much.