In the multi-attribute decision making (MADM) studies, decision making is generally performed based on the expert judgments and it brings a risk of subjectivity and bias in the evaluation of the decision process. In this study, we propose to combine the decisions based on expert judgment and data knowledge to reduce those risks. A new approach is introduced to define the decisions based on data knowledge with fuzzy random variables in a similar manner of the decisions based on expert judgment. Since these fuzzy linguistic variables have the same features with the counterparts used in the expert judgment part, we can combine the decisions coming from both parties, expert judgement and data knowledge, easily. After this process, a chosen MADM method can be performed and the alternatives are ranked.