Joule

Baumgärtner, C.L., Way, R., Ives, M.C., Farmer, J.D., (2024), The need for better statistical testing in data-driven energy technology modeling, Joule, https://doi.org/10.1016/j.joule.2024.07.016.

Forecasting the future evolution of clean energy technologies is vital to the energy transition. To develop effective policies and make sound investments, energy technology models need to be reliable, and we need scientifically justified ways to assess their reliability. We analyze the reliability of several data-driven energy technology models from the recent literature, including several “S-curve” models, which have recently gained attention in modeling technology diffusion. Our examination shows that, due to a lack of statistical testing, many such models produce unreliable results,for example, underestimating the future deployment of solar photovoltaics and overestimating the future cost of batteries. We highlight the importance of statistical testing and describe various methods to validate a model’s reliability.