The arXiv paper introduces EasyResearch, a follow-up to the earlier EasyClassifier system. EasyClassifier demonstrated that researchers who do not program can still get honest classification results when the correct evaluation procedure is the default. EasyResearch appears to apply the same principle to other common analytic tasks.
The abstract motivates this extension by noting that most applied studies do not merely classify. They predict a numeric value, forecast a series, or classify a signal. EasyResearch is designed to make honest evaluation the default for these tasks as well, so that non-programming researchers can avoid methodologically weak practices without writing code.
Because the abstract is cut off, the source does not provide details on how EasyResearch implements these defaults or how it handles different data types. The full paper is available on arXiv, and the title suggests it covers regression, forecasting, and signal classification in a unified way.