Auto-generation of Data Exploratory Sessions Based on Deep Reinforcement Learning
Exploratory Data Analysis(EDA)is a data analysis method aimed at revealing the structure,patterns,and relationships in a dataset through visualization and summary statistics.Data analysts can interactively explore unfamiliar datasets through operations and provide users with preliminary insights.Deep Reinforcement Learning(DRL)has been proven to address many difficult Artificial Intelligence challenges.One can attempts to combine the EDA and DRL,proposing a system called AEDAS.The system models EDA as a control decision problem,combining a novel DRL architecture to automatically generate the persuasive exploratory sessions and present them in the form of EDA notebooks.Experiments show that the EDA notebooks generated by the system can provide users with tangible and effective preliminary insights.
exploratory data analysisDeep Reinforcement Learning architecturecontrol problemexploratory sessions
国家科技期刊平台
NSTL
万方数据
