In recent years, the rise of Internet traffic has expanded explosively as a result of the quick increase in the number of Internet users. As a result of the exponential rise in Internet applications and their high computing costs, both port-based strategies and DPIs (deep packet inspections) are less effective. An integral component of the networking domain that transforms traditional networking into an automated network will be software defined networking. SDNs (Software Defined Networks) are used to centralize network architectures. The control planes and data planes have been separated as a result of SDNs. This has also resulted in the creation of a centralised network controller with comprehensive views of complete networks. Therefore, there is only one control plane (SDN controller) for all the switches in SDNs, in contrast to traditional networks where the two levels of control and data are linked together. Data planes are in charge of straightforward data packet forwarding while control planes do traffic routing. One key issue of this new networking architecture is security of data and identification of malicious packets. This paper uses the traffic information dataset of SDNs in an attempt to select most important features required for classifications of network packets into normal and malicious classes. The proposed scheme called ITCFFE (Intelligent Traffic Classification Feature Engineering Technique) is based on correlations between features and BFEs (Backward feature Eliminations) for dropping unwanted features while retaining the most important features for its final outcomes. The proposed ITCFFE schema is evaluated using multiple classifiers for its efficiency where classification accuracy of more than 95% is achieved.