A Machine Learning Web Application for Real-Time Thermoelectric Property Predictions

Abstract

The growing adaptation of machine-learning studies for thermoelectric research is transforming the realm of material discovery. However, many of these studies are purely investigative and offer limited practical utility. Herein, we developed a resource- and cost-effective, single-page application machine-learning framework. The application is deployed in Microsoft Azure cloud using Docker containers, with the static webpage hosted on GitHub. The web page, https://kleinkeresearchgroup.github.io/TE_Predictions, is publicly accessible through a browser-based interface and serves as guidance for researchers to predict thermoelectric properties. In addition to thermoelectric property predictions from previous studies, we include an additional machine learning model for property predictions on the power factor of thermoelectric materials. Moreover, we developed an Azure Function App, https://fn-research-kleinke-app-jl-2025.azurewebsites.net/index, as an alternative to present a rapid and affordable option for deployment.