Analysis of Learning Time Data using Generalized Linear Bayesian Item Response Theory

In this study, we analyzed the AI E-learning course data from October 2023 to February 2024 (245 participants, 28 training courses). The model used is based on Bayesian item response theory with Exponential and Weibull, Inverse Gaussian distributions in order to analyze the trend of learning time for each training course and to measure the three levels of learning time deviation, learning laziness and learning promotion. The model also allows for analysis with small amounts of data through Bayesian modelling. From the results, the learning status of each training course can be visualized as a two-dimensional plot or histogram, and the model is expected to be operated as a system development for visualizing training and education data. The model maintains high prediction accuracy even with real-world data.