Cybercrimes committed using computer networks lead to billions of dollars lose, the illegal access into computer system, stealing valuable data and destroying organization networks which in turn affect the cyber resources. Because of the expansion of attacks or threats on the networks infrastructure, which is nothing but can be consider as an illegitimate intrusion, based on the machine learning methodology, the intrusion detection system (IDS) can consider as one of the most used cyber security mechanisms, thus to detect the promiscuous activities against sensitive and private data. In this paper our target is to provide a guide lines for researchers and developers discussing the IDS construction phases and their latest techniques, we will clarify the most applied data sources employed in the proposition of a model that will be built for the purpose of creating an intelligent detection system. Furthermore, this survey presents the most commons and latest methods employed and used for designing an IDS based on the data mining techniques and discusses the artifacts removal by summarizing the advantages with the disadvantages of the currents methods and addressing the last novel steps into this field of research.