R.ROSETTA: an interpretable machine learning framework
چکیده(به انگلیسی؛ چکیدهی فارسی در دسترس نیست)
Abstract Background Machine learning involves strategies and algorithms that may assist bioinformatics analyses in terms of data mining knowledge discovery. In several applications, viz. Life Sciences, it is often more important to understand how a prediction was obtained rather than knowing what made. To this end so-called interpretable machine has been recently advocated. study, we implemented an package based on the rough set theory. An aim our work provision statistical properties models their components. Results We present R.ROSETTA package, which R wrapper ROSETTA framework. The original functions have improved adapted programming environment. allows for building analyzing non-linear models. gathers combinatorial statistics via rule-based modelling accessible transparent results, well-suited adoption within greater scientific community. also provides visualization tools facilitate minimization analysis bias noise. freely available at https://github.com/komorowskilab/R.ROSETTA . illustrate usage applied transcriptome dataset from autism case–control study. Our tool provided hypotheses potential co-predictive mechanisms among features discerned phenotype classes. These co-predictors represented neurodevelopmental autism-related genes. Conclusions new insights knowledge-based systems. demonstrated facilitated detection dependencies Although sample application illustrates analysis, can be used analyze any organized decision tables.