The significance of gas compressibility factor in petroleum engineering encourages the researchers to employ the most accurate and precise methods for estimation of this factor. Commonly, empirical correlations due to their simplicity have been referred more than other approaches for prediction of Z-factor. There is no clear and reliable report to address an appropriate combination of correlation and mixing rule for each type of gas. In the present study, combination of several empirical correlations and mixing rules is examined and a decision tree is constructed to suggest best combination for each gas system. For this reason, 2329 experimental data were used for analysis. According to the results, Leland–Mueller mixing rule/Sanjari and Lay correlation is the best combination for sour and natural gas. Also, Van Ness–Abbot mixing rule/Hall–Yarborough correlation, Stewart–Burkhardt–Voo mixing rule/Heidarian correlation and Satter–Campbell mixing rule/Papay correlation are the most appropriate combination for gas condensate, binary and ternary mixtures respectively. For binary mixtures, a robust and novel empirical correlation was developed based on Kay mixing rule to estimate Z-factor. The results employed how good the new correlation is in agreement with the experimental data with significant R-squared 0.9843.