Towards green and sustainable artificial intelligence: quantifying the energy footprint of logistic regression and decision tree algorithms

Artificial Intelligence (AI) significantly impacts the advancement in numerous application domains such as healthcare, education, energy systems, transportation, entertainment, weather forecasting, and finance. However, the exponential increase in the use and adoption of AI algorithms, particularly Machine Learning (ML) algorithms, has raised serious concerns about their environmental impact due to their higher computational demand. In the literature, most of the research in AI has focused on increasing the performance of the algorithms without considering their effects on energy consumption. In this context, this research aims to address the need to quantify the energy footprint of machine learning algorithms to pave the way towards green and sustainable AI. This paper conducted an experimental study on the energy consumption of two state-of-the-art machine learning algorithms – logistic regression and decision tree while applying them to a popular benchmark Iris dataset. The energy consumption is quantified and compared through thorough measurement and analysis. The findings of this research contribute to the ongoing efforts to enable eco-conscious and sustainable AI applications and systems. The paper source code is available online on Github1 .