Interaction system for helping user to construct machine learning model

A method includes: (a) receiving a training dataset, a test dataset, a number of iterations, and a parameter space defining possible parameter values of a base model; (b) for the number of iterations, performing a parameter search process that produces a report including information about the plurality of machine learning models, where the parameter search process includes: (i) generating a Bayesian optimized parameter space having options for verification by hierarchical Kfold cross-validation, and (ii) generating a Bayesian optimized parameter space having options for verification by hierarchical Kfold cross-validation; wherein the optimized parameter set comprises training data from the training data set and test data from the test data set; (ii) running the base model with the final optimized parameter set, thereby producing model results of the plurality of machine learning models; (iii) calculating a Kolmogorov-Smirnov (KS) statistical magnitude of a model result; and (iv) saving the model results and t