<p>With the wide access to data and advanced technologies, organizations and firms prefer to use data-based and interpretable analytics to deal with uncertain and cognitive decision-making problems. In this regard, this study considers quantitative data and qualitative variables, to propose a multi-dimensional decision framework based on the nested probabilistic linguistic term sets. Under the framework, XGBoost algorithm, one of the machine learning methods, is conducted to capture the importance of attributes by using the historical data, and further calculate the attribute weights. The constrained parametric approach is used to establish membership functions of linguistic variables, and then get the objective probabilities in the linguistic model, so that we can obtain a scientific decision matrix. A case study concerning the ranking of bank credit is applied to present the proposed decision framework, and the process of making a rational decision. According to the comparative analysis, the proposed framework is flexible and the result is stable. Managers and policymakers determine the attribute weights by real data and choose the suitable decision method for a certain application. The framework provides an opportunity for capturing, integrating, analyzing data, and interpreting linguistic variables in the model to consider uncertain and cognitive decision at the both theoretical and practical levels. </p>